1 INFO-VAX	Fri, 28 Feb 2003	Volume 2003 : Issue 116       Contents:5 ?= Javadev118 and Java 1.4 on same OVMS 7.3-1 system. 1 Alpha CONDIST or Alpha VMS 7.3 - 7.2... in the UK 0 Re: Anybody know why HP didn't get the business?6 Re: Anyone use a Compaq 20/40 dlt tape drive in a vax?. Bill Gates says you "get what you pay for" ...9 Re: EMC to acquire Legato ?  How about OpenVMS + Legato ? 8 Re: EMC to acquire Legato ? How about OpenVMS + Legato ? HP - crimes do pay after allH Re: IBM says AMD dead in 5yrs ... -- Microsoft Monopoly vs. IBM monopolyD Re: Idea for installing VMS without a 512byte CD-ROM...Will it work?D Re: Idea for installing VMS without a 512byte CD-ROM...Will it work?D Re: Idea for installing VMS without a 512byte CD-ROM...Will it work?D Re: Idea for installing VMS without a 512byte CD-ROM...Will it work?D Re: Idea for installing VMS without a 512byte CD-ROM...Will it work?5 Re: Itanium "threw out all the good parts of the x86" / Re: Linus trashes Itanium - he's scared of VMS! / Re: Linus trashes Itanium - he's scared of VMS! , RE: More Tar problems (was  VMSTAR question), RE: More Tar problems (was  VMSTAR question)! Re: Official OpenVMS name for the P Re: OpenVMS.org: Marvel article and HP's press release for Marveland  Alpha RetaP Re: OpenVMS.org: Marvel article and HP's press release for Marveland Alpha RetaiP Re: OT but worth posting, virus warning: "FWD: Look at this pack	from Microsoft.3 Raise CPUMAXIMUM (CPU time limit) for a running job 7 Re: Raise CPUMAXIMUM (CPU time limit) for a running job 7 Re: Raise CPUMAXIMUM (CPU time limit) for a running job 7 Re: Raise CPUMAXIMUM (CPU time limit) for a running job 7 Re: Raise CPUMAXIMUM (CPU time limit) for a running job 7 Re: Raise CPUMAXIMUM (CPU time limit) for a running job  re CD copy booting off disk  Re: Sun and Java+ Re: Suspected compromise of my home VMS box  SW500 - a hardware question  Re: SW500 - a hardware question  Re: SW500 - a hardware question  Re: SW500 - a hardware question  Re: SW500 - a hardware question  RE: SW500 - a hardware question  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: Upcoming VMS improvements ?  Re: VAX Emulators  RE: VAX Emulators  Re: VAX Emulators  Re: VAX Emulators  Re: VMS Backup solutions Re: Volume set across RAID5  Re: Volume set across RAID5  Re: Volume set across RAID5  Re: Volume set across RAID5  Re: Volume set across RAID5  Re: Volume set across RAID5 > Re: What makes the VMS port to Itanium/IPF/whatever difficult?> Re: What makes the VMS port to Itanium/IPF/whatever difficult?1 Re: Wow - OpenVMS Job Advert in UK Computer Press 1 Re: Wow - OpenVMS Job Advert in UK Computer Press   F ----------------------------------------------------------------------  % Date: Fri, 28 Feb 2003 10:46:26 +0100 . From: "Hans M. Aus" <aus@vim.uni-wuerzburg.de>> Subject: ?= Javadev118 and Java 1.4 on same OVMS 7.3-1 system.B Message-ID: <aus-D09A18.10462628022003@wrzx08.rz.uni-wuerzburg.de>  - 1) Why is Javadev118 on my OVMS 7.3-1 system? 7 2) Can I install JAVA 1.4 with out removing Javedev118?   ' Currently DCL $java -fullversion shows:   8 dka100:[sys0.syscommon.][sysexe]java.exe;1 full version  "javadev118:04/03/2000-12:43"   @ $product show history does not list the Javadev118 installation.  6 The OVMS 7.3-1 was a new installation in October 2002.   --  B Cheers, Hans M. Aus, Wuerzburg, Germany,  aus@vim.uni-wuerzburg.de   ------------------------------    Date: 28 Feb 2003 05:31:55 -0800/ From: blokey@mac.com (Edward Blackburn - Stout) : Subject: Alpha CONDIST or Alpha VMS 7.3 - 7.2... in the UK= Message-ID: <467f78fc.0302280531.25f4b66a@posting.google.com>   F I am in the process of replacing my hobbyist VAXen with an Alpha (DS10F from islandco.com!) and was wondering if anyone in the UK would have a0 copy of the CONDIST for me to beg/borrow/buy....  F I would get the Montagar CDs but they are *sold out* of the Alpha CDs.  F Oh, and if anyone wants a couple of VAXstation 4000/90As with 32MB RAMA and a couple of 1.05GB HDDs (pickup from Bridgend - Wales - UK or  post) let me know.   Thanks in advance.   Ed.    -- Edward Blackburn - Stout Systems Programmer at large  blokey at mac dot com    ------------------------------  # Date: Fri, 28 Feb 2003 14:30:58 GMT # From: "John Smith" <a@nonymous.com> 9 Subject: Re: Anybody know why HP didn't get the business? F Message-ID: <CoK7a.11462$em1.576@news04.bloor.is.net.cable.rogers.com>  < "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message! news:3E5EDF86.74022B3F@fsi.net...  > John Smith wrote:  > > 8 > > ARCHIPELAGO PICKS SUN SERVERS FOR OTC STOCK TRADING,& > > REPLACING MICROSOFT/INTEL MACHINES > > A > > In a move to cut costs, boost speed, and increase its trading  volume,  > > Archipelago : > > is shifting its OTC stock-trading business over to Sun Microsystems > > Unix-based% > > servers by the end of next month.  > E > Did HP even try for it (we all know how wonderful OpenVMS marketing 	 > is...)?     + They could have pitched Tru64 or PH-UX too.    ------------------------------   Date: 28 Feb 2003 12:04 CDT ' From: carl@gerg.tamu.edu (Carl Perkins) ? Subject: Re: Anyone use a Compaq 20/40 dlt tape drive in a vax? - Message-ID: <28FEB200312043281@gerg.tamu.edu>    jeremy@vsm.com.au writes...  }Jim Agnew wrote: K }> Version 5.5-2 VMS  they say it's not supported til v7 or so, but if it's I }> a generic scsi tape, that works, that's enuf for me.. I just happen to F }> have one laying around, and would like to see if it works before we! }> cought up for an adapter.. ;-)  }>   }> JimO }I believe the highest-capacity DAT drive supported on OpenVMS is 12/24GB (e.g. H }TLZ10).  Apparently the 20/40 drives don't support the writing of "odd"M }blocks, i.e. tape blocks with an odd number of bytes, and this functionality $ }is assumed by the BACKUP utility.  ) }Of course, you may find it works anyway.   < My XP900 has a Compaq SDT-10000 DAT drive, which is 20/40GB.5 It works fine under V7.2-1. (It's also quite speedy.)   G It is a supported option. See the first thing (and some later ones) on: ? http://h18000.www1.hp.com/emea/presalessupport/SupportInfo.html   H It isn't supported on versions as old as 5.5-2, but it has a fair chanceF of working. You may have to watch out for mismatched speeds - the tapeF drive may be faster than your disks if they are of anything resemblingE the same vintage as VMX V5.5-2 (a 20/40 DAT drive does 2.4MB/sec when H in the "20" mode, i.e. without compaction, and 4.8MB/sec with compactionJ enabled if you are getting 2:1; a modern hard drive can maintain this evenJ if there is some other activity to the drive, a "vintage" drive might haveB trouble keeping up even if there isn't any other activity at all).   --- Carl   ------------------------------    Date: 28 Feb 2003 08:20:28 -0800( From: bob@instantwhip.com (Bob Ceculski)7 Subject: Bill Gates says you "get what you pay for" ... = Message-ID: <d7791aa1.0302280820.26c37730@posting.google.com>    agrees with me ...  ( http://www.theinquirer.net/?article=8044   ------------------------------  # Date: Fri, 28 Feb 2003 13:33:00 GMT + From: Jeff Cameron <JCam90502@jcameron.com> B Subject: Re: EMC to acquire Legato ?  How about OpenVMS + Legato ?2 Message-ID: <BA84A50C.5227%JCam90502@jcameron.com>   On 2/27/03 8:23 AM, in articleC 20030227162336.46425.qmail@web20203.mail.yahoo.com, "Fabio Cardoso" ! <fabiopenvms@yahoo.com.br> wrote:    > Click  > 6 > http://news.com.com/2100-1001-986230.html?tag=fd_top > 	 > Regards  >  > FC   >  > =====  > ========================== > F?bio dos Santos Cardoso > OpenVMS System Manager > Rio de Janeiro - Brazil  > fabiopenvms@yahoo.com.br > ========================== > 4 > __________________________________________________ > Do you Yahoo!?4 > Yahoo! Tax Center - forms, calculators, tips, more > http://taxes.yahoo.com/   E The biggest problem with Legato and VMS is that it is based on an old E product called WUMPSWARE, which is only a VMS CLIENT. That means your J backups are done on a non VMS Media server, resulting in tape volumes thatJ are not VMS ANSI BACKUP tapes, which means, you cant do native restores toL your VMS system, and although the do provide a plan for disaster recovery ofL your VMS system it is long and drawn out with many steps, of which the first< step is "You rebuild your VMS system disk from distribution.  F The same goes for Veritas Netbackup for VMS. It is also a client only.  - We are resellers for both Legato and Veritas.    Jeff Cameron MTI    ------------------------------  % Date: Fri, 28 Feb 2003 09:15:07 -0500  From: norm.raphael@metso.comA Subject: Re: EMC to acquire Legato ? How about OpenVMS + Legato ? ? Message-ID: <OF95B80401.61281433-ON85256CDB.004E2FAD@metso.com>   = Why are you ignoring the Legato OpenVMS Storage Node product?   C From:  Jeff Cameron <JCam90502@jcameron.com> on 02/28/2003 08:33 AM   7 Please respond to Jeff Cameron <JCam90502@jcameron.com>    To:    Info-VAX@Mvb.Saic.Com cc:   E Subject:    Re: EMC to acquire Legato ?  How about OpenVMS + Legato ?      On 2/27/03 8:23 AM, in articleC 20030227162336.46425.qmail@web20203.mail.yahoo.com, "Fabio Cardoso" ! <fabiopenvms@yahoo.com.br> wrote:    > Click  > 6 > http://news.com.com/2100-1001-986230.html?tag=fd_top > 	 > Regards  >  > FC >  > =====  > ========================== > F?bio dos Santos Cardoso > OpenVMS System Manager > Rio de Janeiro - Brazil  > fabiopenvms@yahoo.com.br > ========================== > 4 > __________________________________________________ > Do you Yahoo!?4 > Yahoo! Tax Center - forms, calculators, tips, more > http://taxes.yahoo.com/   E The biggest problem with Legato and VMS is that it is based on an old E product called WUMPSWARE, which is only a VMS CLIENT. That means your J backups are done on a non VMS Media server, resulting in tape volumes thatJ are not VMS ANSI BACKUP tapes, which means, you cant do native restores toI your VMS system, and although the do provide a plan for disaster recovery  ofF your VMS system it is long and drawn out with many steps, of which the first < step is "You rebuild your VMS system disk from distribution.  F The same goes for Veritas Netbackup for VMS. It is also a client only.  - We are resellers for both Legato and Veritas.    Jeff Cameron MTI    ------------------------------  # Date: Fri, 28 Feb 2003 15:19:14 GMT # From: "John Smith" <a@nonymous.com> % Subject: HP - crimes do pay after all H Message-ID: <S5L7a.12202$em1.10936@news04.bloor.is.net.cable.rogers.com>  F http://story.news.yahoo.com/news?tmpl=story2&cid=569&ncid=738&e=5&u=/n' m/20030228/tc_nm/tech_hewlettpackard_dc     1 HP CEO Got 2002 Bonus, Capellas Left with $26 Mln = Thu Feb 27, 9:30 PM ET  Add Technology - Reuters to My Yahoo!   ? LOS ANGELES (Reuters) - Carly Fiorina, chief executive of No. 1 E personal computer and printer maker Hewlett-Packard Co (HPQ.N), got a C pay package worth more than $10 million last year, while lieutenant @ Michael Capellas received more than $26 million after resigning, filings showed on Thursday.   @ Fiorina, the main engineer of the acquisition of Compaq ComputerF Corp., earned a $1 million salary in 2002 and received a bonus of $2.9C million, a raise after receiving no bonus in 2001, voting materials @ sent to shareholders ahead of the April 2 annual meeting showed.  D Fiorina also took home stock options valued at $6.8 million in 2002,@ although at HP's current stock price they could not be exercised profitably.   C The HP board also raised the maximum bonus Fiorina could receive to : $12 million from $9 million annually, citing her increased responsibilities.   E Fiorina's compensation became a subject of controversy last year when E former director Walter Hewlett said the board was considering raising E Fiorina's salary to $70 million over two years upon the completion of  the controversial merger.   A Fiorina put together the merger with the help of Capellas, former E Chief Executive of Compaq, and both declined retention bonuses during 	 the deal.   A Capellas' Compaq contract assured him compensation if he left the & company after losing his title of CEO.  ? Capellas left HP last year and took over as Chief Executive and E Chairman of bankrupt telecoms company WorldCom Inc., where he will be C paid as much as $5 million in salary and bonuses and $18 million in  restricted stock.   D Capellas' package for HP last year included $800,000 in salary, $1.9> million in performance bonus and a separation payment of $14.46 million. HP gave him $9.6 million more to cover taxes.  E HP on Tuesday posted a fiscal first-quarter profit at the high end of E expectations, driven by cost cutting and a narrower operating loss in D its enterprise computing business, but revenue of $17.9 billion fell short of Wall Street forecasts.   A That prompted some analysts to question whether the merger, which D combined Compaq's large and loss-making PC business and its computerC server business with HP's offerings in those areas, was working. HP D also has a market-leading and extremely profitable printer business.  It also sells computer services.   -=-----   F VMS marketing and advertising would have made better use of the money.   ------------------------------  ! Date: Fri, 28 Feb 03 12:00:20 GMT  From: jmfbahciv@aol.com Q Subject: Re: IBM says AMD dead in 5yrs ... -- Microsoft Monopoly vs. IBM monopoly + Message-ID: <b3nlil$20t$2@bob.news.rcn.net>   , In article <3E5EB9A6.26BC1B00@computer.org>,0    Sam Yorko <JOATnospamMON@computer.org> wrote: >jmfbahciv@aol.com wrote:  >>  5 >> I guess I should point out that I'm biased.  :-)))  >>   >> /BAH  >  >You are?  ;-)  > When I have my development and maintenance hats on, of course.& When I had my stock holder hat on, no.   /BAH  ' Subtract a hundred and four for e-mail.    ------------------------------    Date: 28 Feb 2003 02:33:37 -0800/ From: robl@no-no-badpuppy.com (Robert Lawrence) M Subject: Re: Idea for installing VMS without a 512byte CD-ROM...Will it work? = Message-ID: <eebf2e9b.0302280233.4a222912@posting.google.com>   ` "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3E5EDB1D.5D8A37ED@fsi.net>... > "Mark E. Levy" wrote:  > > < > > "John Laird" <john@laird-towers.org.uk> wrote in message6 > > news:cees5vc2p0btigm5i4ucoq75nt0uk2e5kv@4ax.com...J > > > On Thu, 27 Feb 2003 15:41:50 GMT, "Mark E. Levy" <mlevy70@attbi.com>	 >  wrote:  > > >  > > > >> Robert Lawrence wrote:  > > > >>  > M > > > >> > Will this work?  If I dump the CD contents byte for byte to a SCSI N > > > >> > hard drive on a PC, will the VAX be able to read and boot from that > > > >> > drive?  > > > > M > > > >No, because you won't be writing a valid boot block on the SCSI drive.  > > > N > > > Uh ?  I don't know why there is quite so much suspicion about this idea.H > > > It'd be the first thing I'd try if I had just these tools to hand. > > N > > Because you can't guarantee exactly where the data will end up on the hardN > > drive. The bootblock must be in a specific place. I do agree, though, that@ > > there is no harm in trying it. Just don't get your hopes up. > > O > > > If the CD is bootable, a block-for-block copy of it to a hard disk should P > > > retain that capability.  It will look odd when mounted as it will have theK > > > volume size, cluster factor, etc, of a CDROM, but this shouldn't be a  > > > showstopper. > >  > > It could be. > " > Following onto what Mark said... > G > Think about this: you're laying an image of a disk no more than 650MB D > (o.k., maybe 700) onto a device with considerably larger capacity. > 1 > Does anyone see a problem with that besides me?   F Hrmmm.  But...this hard drive would only have the CD image long enoughE to perform an OS install.  Once that is completed, the drive would be # initialized as a normal data drive.   4 The idea was to be a quick and dirty temporary hack.   ------------------------------  % Date: Fri, 28 Feb 2003 14:49:00 +0000 + From: John Laird <john@laird-towers.org.uk> M Subject: Re: Idea for installing VMS without a 512byte CD-ROM...Will it work? 8 Message-ID: <73tu5vcvmm07hqevjlm06ks34m9ln2g9v3@4ax.com>  , On Thu, 27 Feb 2003 18:00:12 -0400, JF Mezei( <jfmezei.spamnot@vl.videotron.ca> wrote:   >John Laird wrote:L >> If the CD is bootable, a block-for-block copy of it to a hard disk shouldM >> retain that capability.  It will look odd when mounted as it will have the H >> volume size, cluster factor, etc, of a CDROM, but this shouldn't be a >> showstopper.  > M >Different driver involved CD vs DISK. You make an unverified assumption that L >the ROM drivers that manage the boot, and subsequent VMS drivers will treat >the disk drive as a CD-ROM.  I You seem to be suggesting that the rom drivers will mount up a hard disk, L find some volume information that suggests it is otherwise, and then attemptK from that point to treat the drive as a cdrom ?  Sounds pretty far-fetched.   O >Having said that, I am not 100% sure, but I think that I did manage to boot my H >all mighty Microvax II from CD (at least Stabackup) once. Its CD-ROM isH >connected to a Dilog Qbus controller which emulates MSCP, so the CD-ROM  >appears to VMS as a dick drive. > F >This would mean that perhaps the placement of data on the CD would beM >compatible, but it would have to be a High Sierra format (natice VMS format)  >and not an ISO-9660 one.   H You seem to be arguing my case.  (I certainly make no claims for non-ODSJ file structures, mind.)  The key thing is that both device types are blockL structured, with 512-byte blocks.  Provided the mapping of logical blocks toJ physical blocks does not change, I fail to see why file systems should not( be interchanged between different media.  N >HELP DUMP /DESCRIPTOR seems to indicate that ISO 9660 is treated differently.  	 No doubt.      	John    ------------------------------  % Date: Fri, 28 Feb 2003 14:53:37 +0000 + From: John Laird <john@laird-towers.org.uk> M Subject: Re: Idea for installing VMS without a 512byte CD-ROM...Will it work? 8 Message-ID: <7otu5vcc1kl2ccgfdpmrtrn9mqcjm2j676@4ax.com>  I On Thu, 27 Feb 2003 16:26:30 +0000, John Laird <john@laird-towers.org.uk>  wrote:  L >On Thu, 27 Feb 2003 15:41:50 GMT, "Mark E. Levy" <mlevy70@attbi.com> wrote: >  >>> Robert Lawrence wrote: >>>  >H >>> > Will this work?  If I dump the CD contents byte for byte to a SCSII >>> > hard drive on a PC, will the VAX be able to read and boot from that  >>> > drive? >>H >>No, because you won't be writing a valid boot block on the SCSI drive. > I >Uh ?  I don't know why there is quite so much suspicion about this idea. C >It'd be the first thing I'd try if I had just these tools to hand.   H Well, I can think of one problem - if the distribution media was createdB with only cdrom driver support.  I'd put money on dkdriver (or itsF derivatives) shipping as is, with implicit support for all disk types, though.   I Generally, I subscribe to the view that VMS Engineering will tend to make L things work widely, rather than break at the slightest opportunity.  I know,. I know, unsupported and all that, but still...     	John    ------------------------------  % Date: Fri, 28 Feb 2003 14:58:21 +0000 + From: John Laird <john@laird-towers.org.uk> M Subject: Re: Idea for installing VMS without a 512byte CD-ROM...Will it work? 8 Message-ID: <o2uu5v0576b9m1ck9898jkbfrjpjj1lv7g@4ax.com>  H On 28 Feb 2003 02:33:37 -0800, robl@no-no-badpuppy.com (Robert Lawrence) wrote:  G >Hrmmm.  But...this hard drive would only have the CD image long enough F >to perform an OS install.  Once that is completed, the drive would be$ >initialized as a normal data drive.  L I trust you have a second drive, then ?  The minimum/install environment mayK *not* be inclined to let you initialise and install to the same device that  it has booted from...   5 >The idea was to be a quick and dirty temporary hack.   J Do try it, and report back.  We are all ears, even if some of them seem to be burning ;-)  G Make sure your Unix box is correctly seeing 512-byte sectors on the CD.      	John    ------------------------------  % Date: Fri, 28 Feb 2003 14:37:24 +0000 + From: John Laird <john@laird-towers.org.uk> M Subject: Re: Idea for installing VMS without a 512byte CD-ROM...Will it work? 8 Message-ID: <h6su5vgjkoshd8t406cepup9mv2fq2ajdu@4ax.com>  K On Thu, 27 Feb 2003 20:48:14 GMT, "Mark E. Levy" <mlevy70@attbi.com> wrote:   9 >"John Laird" <john@laird-towers.org.uk> wrote in message 3 >news:cees5vc2p0btigm5i4ucoq75nt0uk2e5kv@4ax.com... G >> On Thu, 27 Feb 2003 15:41:50 GMT, "Mark E. Levy" <mlevy70@attbi.com>  >wrote:  >> >> >> Robert Lawrence wrote: >> >>  >J >> >> > Will this work?  If I dump the CD contents byte for byte to a SCSIK >> >> > hard drive on a PC, will the VAX be able to read and boot from that  >> >> > drive? >> >J >> >No, because you won't be writing a valid boot block on the SCSI drive. >>K >> Uh ?  I don't know why there is quite so much suspicion about this idea. E >> It'd be the first thing I'd try if I had just these tools to hand.  > K >Because you can't guarantee exactly where the data will end up on the hard  >drive.   E I thought "block-for-block copy" was clear enough, but obviously not.   D >The bootblock must be in a specific place. I do agree, though, that= >there is no harm in trying it. Just don't get your hopes up.   H I think the OP mentioned a MV3100.  They don't need a boot-block, as theI boot roms cope just fine with traversing the ODS file structure.  If that K gets shot during the processing of raw-copying, then it won't work.  But if I it is copied consistently, my conjecture is that it will be mountable and A bootable.  (Also, any boot-block would be copied faithfully too.)   L >> If the CD is bootable, a block-for-block copy of it to a hard disk shouldM >> retain that capability.  It will look odd when mounted as it will have the H >> volume size, cluster factor, etc, of a CDROM, but this shouldn't be a >> showstopper.  > 
 >It could be.   I There is ample evidence that a "bootable" VMS disk, created on a suitable L container disk, be that a physical device or an LD or VD container file, canE be transferred to a non-VMS machine and burnt to CD.  That CD will be I bootable.  I have plenty of non-bootable backup disks which I created the L same way.  Not one looked like a CD when it was "formed", but they all mountI just fine.  Equally, I have gone the other way, copying CD disks block by H block to container files and then mounting them as normal disks (I don'tF recall off-hand what device type LDDRIVER sets up, but it sure ain't aJ CDROM).  These too work just fine.  I'll grant there is a possibility thatG MOUNT might object to volume characteristics that don't square with the H physical device characteristics, but, frankly there is no real evidence.  K It's only a shame I don't have a spare disk I could sacrifice to test this.      	John    ------------------------------  # Date: Fri, 28 Feb 2003 15:47:20 GMT 0 From: rw@vorpalbunnyeircom.net (Russell Wallace)> Subject: Re: Itanium "threw out all the good parts of the x86"0 Message-ID: <3e5f8461.183428670@news.eircom.net>  B On 27 Feb 2003 02:27:31 -0800, xeio77@hotmail.com (Tsarkon) wrote:  D >Save one thing - no one buys the IS64 and it sucks a lot worse thanF >IA32., I have owned and used quite a number of systems, and I have toE >say, after some limited exerience with both Itanium1 and 2 machines, : >you would be a fruit not to like just about nything else.  F Cool, someone with hands-on experience! What have you found to dislike@ about Itanium 2? (It's supposed to do well on benchmarks, unlike version 1.)    --  3 "Mercy to the guilty is treachery to the innocent." + Remove killer rodent from address to reply. ! http://www.esatclear.ie/~rwallace    ------------------------------  % Date: Fri, 28 Feb 2003 14:00:13 +0000 ' From: Andrew Harrison SUNUK Consultancy 8 Subject: Re: Linus trashes Itanium - he's scared of VMS!. Message-ID: <3E5F6B6D.7050802@nospamn.sun.com>   Bill Todd wrote:M > "Andrew Harrison SUNUK Consultancy" <Andrew_No.Harrison_No@nospamn.sun.com> ; > wrote in message news:3E5E1E82.8080307@nospamn.sun.com...  >  >> >>Bill Todd wrote: >>@ >>>"JF Mezei" <jfmezei.spamnot@vl.videotron.ca> wrote in message, >>>news:3E5D17C1.EFC0550B@vl.videotron.ca... >>>  >>> , >>>>Andrew Harrison SUNUK Consultancy wrote: >>>> >>>>6 >>>>>IA32 is a speed deamon its design allows it to be8 >>>>>clocked faster and implimented in a smaller process& >>>>>earlier than a brainiac like IPF. >>>>K >>>>Oh my god. The end of the world is close because I find myself agreeing  >>>  >>>with  >>>  >>>  >>>>Andrew Harrisson.  >>>  >>> D >>>I'm afraid Andrew's not worth agreeing with this time.  Itanic2's >> > problem is > B >>>not lack of competitive performance, but lack of efficiency (inG >>>power/required cooling, chip area/manufacturing cost, expenditure of C >>>development funding and effort, requirement for more compilation G >>>optimization and updates thereof...).  And of course the lack of the C >>>ancillary on-chip performance and scaling features that EV7 has.  >>>  >>< >>Interesting in fact you are agreeing with me on your first >>point  >  > J > No, I am not:  read more carefully.  Mosberger's contention that Itanic2L > seems equal and possibly superior to the P4/Xeon in performance at a givenM > process is correct, and your contention that there's something about Itanic G > that prevents it from being implemented in as small a feature size as N > P4/Xeon sounds very much like bullshit (the reason it's not currently in 130J > nm. is far more likely that a pretty much from-the-ground-up design likeJ > Itanic2 is risky enough without targeting your most leading-edge processN > technology for its first implementation, leaving aside the question of whereJ > it makes the most sense for Intel to use its limited 130 nm fab capacityK > right now if it doesn't want to lose its bread-and-butter market to AMD).  > 2 Sorry but this isn't true for a number of reasons.   To start off with E IPF has the largest die size of any currently produced microprocessor > this in itself makes it an unlikely candidate to be first intoE a new process because of initial yield issues. Its die size is highly  influenced by its design.   F Secondly there is a lot you can do when designing a CPU to facilitiateG its move into a newer faster process and vica versa. CPU clocking being ' one major area which can catch you out.   E A classic example and one that caused Sun great pain so I am speaking B from experience here was SuperSPARC. At the time it had one of theG highest inst/cycle counts so that at a given clock rate it outperformed F Alpha etc. SuperSPARCs problem was that we were unable to ramp up its @ clock speed or impliment it in the smaller processes that TI had@ available because of an number of design decisions made when we  implimented the chip.   > The choice of going for speed derived from high clock rates or= speed derived from doing more in each clock cycle is a design : decision, chip architects make these choices and they then8 drive the process/frequency that the CPU is built in and< operates at. This was why Michael Slater coined the Brainiac( Speed Deamon categories for CPU designs.   Regards  Andrew Harrison    ------------------------------  % Date: Fri, 28 Feb 2003 12:37:56 -0500 A From: "Fred Kleinsorge" <kleinsorge@star-dot-zko-dot-dec-dot-com> 8 Subject: Re: Linus trashes Itanium - he's scared of VMS!. Message-ID: <3e5f9e74$1_2@hpb10302.boi.hp.com>  5 "Bill Todd" <billtodd@metrocast.net> wrote in message , news:d6CcnYRy8sYOFcOjXTWcow@metrocast.net... >  > F > I don't like or particularly respect Itanic, largely for the reasons already G > noted:  it achieves its competitive performance by the application of  brute J > force and band-aids to a fairly misbegotten architectural approach.  But the G > resulting raw performance *is* competitive, even though the resulting ; > price/performance and ease of future enhancements is not.  >   G It's price/performance is competetive.  It is priced to compete against L other 64-bit chips used in UNIX servers.  It's "ease of future enhancements"L isn't really interesting - only the results.  And even without architecturalH innovation, it's - as you indicate - "brute force" approach lends itself; well to simply driving the process down and clock rates up.    ------------------------------  % Date: Fri, 28 Feb 2003 08:22:40 -0500 1 From: "Farrell, Michael" <MFarrell@voltdelta.com> 5 Subject: RE: More Tar problems (was  VMSTAR question) P Message-ID: <025766C9BBC5D511A4ED00B0D0F08C23163A26@zny_exchange1.maintech1.com>   Thanks for your response.   G I got a chance to work on it late last night from home by debugging the = source.  I should have realized the problem/situation sooner.   J The file to be tarred is an RMS sequential file of variable length records0 containing a lot of mixed ASCII and binary text.J The VMS2TAR code checks the attributes of the file and knows what to do ifJ the record format is FAB$C_FIX, FAB$C_STM, or FAB$C_STMLF.  It also checks0 the record attributes (fab$b_rat) of the file.    L If it can't make sense of this information, it then assumes that the file isJ a "Unix" file and tries to compute its filesize by reading all the data inL using fgets() and taking the strlen() of the data in the input buffer.  MuchC of the binary data in my records are set to binary zero so the real ' 'bytecount' of the file is compromised.   F The cure for this problem is to change the file attributes of the dataI file(s) to fixed length, 512 byte records and then TAR the files.  ($ SET - FILE/ATTRIB=(LRL:512,RFM:FIX) <FILE NAMES>).    K Then TAR will archive the files properly.  The non-VMS system receiving the I TAR set has to deblock the file(s) when they are extracted to restore its  record characteristics.   
 Thanks again.    Mike Farrell    -----Original Message----- 8 From: 	David J. Dachtera [mailto:djesys.nospam@fsi.net] * Sent:	Thursday, February 27, 2003 11:00 PM To:	Info-VAX@Mvb.Saic.Com 5 Subject:	Re: More Tar problems (was  VMSTAR question)    "Farrell, Michael" wrote:  > G > Thanks for everyone's help yesterday.  We solved the looping problem.  > J > Now I have a new problem.  The files we want to archive in TAR contain a/ > great deal of binary data.  When we TAR them, E > The resulting TAR file is generally less than 1% of the size of the  originalH > file.  (E.G. a 50000 block data file ends up producing a 360 block TARI > file).  Obviously, the data is being lost somehow.  The commands I have  used > are: > 7 > $ tar/create/verbose <tar file name> <data file name>  >  > Or > = > $ tar/create/force/verbose <tar file name> <data file name>  > % > The results are the same with both.  > $ > Any ideas are greatly appreciated.   Well, my questions would be:    F  1. Is 50,000 blocks the EOF (End of file) size or the ALQ (Allocation Quanity) size?  ?  2. Is the file RMS-indexed? (tar will scramble it - CONVERT to " sequential + FDL first, then tar).  F Since tar does no compression, I am at a loss for another explanation, also.    --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Fri, 28 Feb 2003 06:31:10 -0800 1 From: "Farrell, Michael" <MFarrell@voltdelta.com> 5 Subject: RE: More Tar problems (was  VMSTAR question) P Message-ID: <025766C9BBC5D511A4ED00B0D0F08C23163A28@zny_exchange1.maintech1.com>  - I'm resending it after the mailer bounced it.    Sorry for the duplication.   Mike Farrell    -----Original Message-----  From: 	Farrell, Michael   ' Sent:	Friday, February 28, 2003 8:23 AM . To:	'David J. Dachtera'; Info-VAX@Mvb.Saic.Com5 Subject:	RE: More Tar problems (was  VMSTAR question)    Thanks for your response.   G I got a chance to work on it late last night from home by debugging the = source.  I should have realized the problem/situation sooner.   J The file to be tarred is an RMS sequential file of variable length records0 containing a lot of mixed ASCII and binary text.J The VMS2TAR code checks the attributes of the file and knows what to do ifJ the record format is FAB$C_FIX, FAB$C_STM, or FAB$C_STMLF.  It also checks0 the record attributes (fab$b_rat) of the file.    L If it can't make sense of this information, it then assumes that the file isJ a "Unix" file and tries to compute its filesize by reading all the data inL using fgets() and taking the strlen() of the data in the input buffer.  MuchC of the binary data in my records are set to binary zero so the real ' 'bytecount' of the file is compromised.   F The cure for this problem is to change the file attributes of the dataI file(s) to fixed length, 512 byte records and then TAR the files.  ($ SET - FILE/ATTRIB=(LRL:512,RFM:FIX) <FILE NAMES>).    K Then TAR will archive the files properly.  The non-VMS system receiving the I TAR set has to deblock the file(s) when they are extracted to restore its  record characteristics.   
 Thanks again.    Mike Farrell    -----Original Message----- 8 From: 	David J. Dachtera [mailto:djesys.nospam@fsi.net] * Sent:	Thursday, February 27, 2003 11:00 PM To:	Info-VAX@Mvb.Saic.Com 5 Subject:	Re: More Tar problems (was  VMSTAR question)    "Farrell, Michael" wrote:  > G > Thanks for everyone's help yesterday.  We solved the looping problem.  > J > Now I have a new problem.  The files we want to archive in TAR contain a/ > great deal of binary data.  When we TAR them, E > The resulting TAR file is generally less than 1% of the size of the  originalH > file.  (E.G. a 50000 block data file ends up producing a 360 block TARI > file).  Obviously, the data is being lost somehow.  The commands I have  used > are: > 7 > $ tar/create/verbose <tar file name> <data file name>  >  > Or > = > $ tar/create/force/verbose <tar file name> <data file name>  > % > The results are the same with both.  > $ > Any ideas are greatly appreciated.   Well, my questions would be:    F  1. Is 50,000 blocks the EOF (End of file) size or the ALQ (Allocation Quanity) size?  ?  2. Is the file RMS-indexed? (tar will scramble it - CONVERT to " sequential + FDL first, then tar).  F Since tar does no compression, I am at a loss for another explanation, also.    --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Fri, 28 Feb 2003 14:29:59 GMT # From: "John Smith" <a@nonymous.com> * Subject: Re: Official OpenVMS name for theH Message-ID: <HnK7a.11445$em1.10467@news04.bloor.is.net.cable.rogers.com>  ' ""Alan Winston - SSRL Admin Cmptg Mgr"" 1 <winston@SSRL.SLAC.STANFORD.EDU> wrote in message 0 news:00A1C207.4A0A07C0@SSRL.SLAC.STANFORD.EDU...: > In article <3E5E323B.8770E201@vl.videotron.ca>, JF Mezei) <jfmezei.spamnot@vl.videotron.ca> writes:  > >Robert Deininger wrote:A > >> The customers might like some way to distinguish between the  various flavors. > > E > >Then let the customer call it what THEY want. There is no need for 
 Digital/HPD > >to officially give VMS different names. VMS is VMS, especially if the code= > >base will be identical between Alpha and that other thing.  Identical products, E > >but different names ? You'll get plenty "how difficult is the port  from the/ > >real VMS to that industry standard system ?"  > >  > E > If the names are exactly the same, customers won't know which media  kit to > order.    D Microsoft has no problems using someone else's trademark in the name of one of their products:    Microsoft Office v. X for Mac  Microsoft Office 2001 for Mac 6 http://www.microsoft.com/office/howtobuy/macintosh.asp  ? If they're smart they'll soon have Microsoft Office for Lindows    ------------------------------  % Date: Fri, 28 Feb 2003 08:34:43 -0500 & From: jlsue <jlsuexxxz@screaminet.com>Y Subject: Re: OpenVMS.org: Marvel article and HP's press release for Marveland  Alpha Reta 8 Message-ID: <olou5vkuu8q321kjcbhcf9isr61njsfqhd@4ax.com>  , On Thu, 27 Feb 2003 11:58:20 -0400, JF Mezei( <jfmezei.spamnot@vl.videotron.ca> wrote:  
 >jlsue wrote: G >> Is it that "there are certainly customers who have GS160/320 servers = >> who are very happy with the performance of their systems?"  > C >I think that the wildfire stuff is probably similar. Those who had O >promises/expectations set high enough would be disapointed. But those who came % >without expectations would be happy.  > M >If someone purchased the 340 with full knowledge of its slower-than-expected N >cruise speed, he is more likley to be happy. If someone purchased it prior toL >it being produced, based on promises made by the vendor, then if the vendorF >didn't deliver on certain key metrics, the customer would be unhappy.  F So what's your point.  What you say is a truism that can be applied toB almost everything in life, not just computing platforms.  I go buyE Clorox Bleach 'cause the commercials say it'll make my life better... C well, I still gotta put up with the fud, so my expectations weren't E met.  Of course, if I'd done my homework - e.g., read the label - I'd 0 have had my expectations adjusted appropriately.  E And *anybody* who's been around for more than a couple of years knows @ that pre-rease specs can sometimes be wrong.  In my experiences,? though, DEC tended to be more on the conservative side in their ? estimates than most other vendors.  But regardless of all that, B everyone knows that not all applications behave the same, and some: will run faster on Sun, and some will run faster on Alpha.  E The fact is, everybody, and that includes Sun, hype their technology. F And quite often sales folks will overstate things - and I can tell youF from experience (as a customer in the '80s & early '90s) that some Sun3 sales people would say *anything* to make a sale.     @ So what?  The fact is that there is very little difference amongC vendors when you look at the grand scale, and harping constantly on A the minor glitches in each one's camps (e.g., numa issues vs zinc + whiskers) really just wastes a lot of time.   C I'd much rather see more discussions on VMS and VMSclusters in this  group that all this crud.    ------------------------------  % Date: Fri, 28 Feb 2003 08:56:44 -0500 & From: jlsue <jlsuexxxz@screaminet.com>Y Subject: Re: OpenVMS.org: Marvel article and HP's press release for Marveland Alpha Retai 8 Message-ID: <cdpu5vgqev47o61br7ubhohfu2t2vsgcq3@4ax.com>  E On Thu, 27 Feb 2003 16:55:28 +0000, Andrew Harrison SUNUK Consultancy . <Andrew_No.Harrison_No@nospamn.sun.com> wrote:   >  > 
 >jlsue wrote: H >> On Wed, 26 Feb 2003 14:01:13 +0000, Andrew Harrison SUNUK Consultancy1 >> <Andrew_No.Harrison_No@nospamn.sun.com> wrote:  >>     >>  @ >> What?  Is this claim so extraordinary that it requires proof? >>   > 9 >Yes because your own engineers say that there are issues 7 >your own benchmark results illustrate these issues and @ >customers on this newsgroup have benchmarked their applications$ >and discovered them for themselves.  B I see, so now you expect some documentation that I have never seen? something.  Wow, your argument must be really weak to make such  demands.  D But you keep overstating the benchmarks and the engineers statementsE by referencing them out of context.  So far, I've only seen that some E applications of specific types have been found to be more affected by  the numa effect than others.  E Do you really think that's news?  Heck, numa isn't the only computing @ architecture that's suffered under *some* applications.  And theA effect it causes isn't always a binary issue - either it's bad or C good.  As in other things in life, the effects will vary along some D sliding scale depending on the applications - from unnoticeable, allF the way to unusable.  In the end, each business has to decide where on< the scale it is for them.  That's one of the services that a' consultant helps businesses understand.    > > >At the moment the score is 5 for the argument that GS160/320sF >do in fact have performance issues because of their NUMA architecture >and 0 against.  > = >Unless that is you can substantiate your claim to never have : >seen any issues with results from the benchmarks that you= >apparently advise your customers to do the score will remain 
 >as it is.  > Again with this weak argument.  Do you really need to grasp soE desperately at straws?  You can't possibly see that you've overstated D the issues just a bit and back off?  Are you so completely incapable/ of seeing the weaknesses in your own arguments?    > F >> Your statements are more of a general "it's slower than gs140" as aH >> way to warn folks away from it.  You state implicitly (and iirc, evenI >> explicitly at times) that it's  That does qualify, imho, as an example I >> of an extraordinary claim; therefore I ask for proof that this is true H >> in a general sense.  And so far you've only proven that it's true forG >> *some* applications.  If you can't support your claim that it's true > >> in general, then withdraw it, or at least re-qualify it.    >>   > : >Hang on you are the parson who claims that you have never8 >seen an issue. I have simply pointed out that there are9 >customers and benchmark results that show that there are  >issues.  < Now you are telling a big lie.  You have not "simply stated"B *anything*, but have consistently claimed - hell, even hyped - theF fact that the numa effect is a huge problem.  You *never* claimed someC applications and instead implied it is a general-case problem.  You F have made this implication in many ways, for example just in the sheerD number of notes where you discuss this problem, and also in the toneD you use to harp on-and-on about it.  In this very note in fact.  See@ your first paragraph quoted at the top of this reply for a prime example.   > " >Yours is the extraordinary claim. > 8 >So how about answering this question. Did you know that8 >your own engineering teams were advising your customers7 >that ther were issues transitioning from UMA to NUMA ?   E I have known that some applications experience problems from the Numa C effect, yes.  I have also worked with applications that do not.  So E what?  Numa issues (in general) are not news, and are somewhat public A knowledge.  Anyone who has been in computing for more than, say 5 @ years, should have some more detailed understanding of computing? technology.  Anyone with less time than that shouldn't have the C responsibility to make these blunders.  But I'm sure that there are D some that do anyway, and sometimes it's even Sun products that don't meet their expectations.   > 9 >And if you did know what was the basis of your engineers 	 >advice ?   F Do I have to know this?  Why is this a requirement for myself, but notA for you?  Do YOU know what was the basis of their advise?  Please  provide these details.    C Do you know under what conditions the numa effect becomes an issue? B And to what extent it affects the businesses?  Can you tell me theA business value of the effects?  Do you have some inside info that & proves it's true for all applications?   > 3 >This shouldn't be too difficult for you to answer.  >    Ditto.   ------------------------------  # Date: Fri, 28 Feb 2003 13:46:06 GMT 1 From: Michael Austin <maustin@firstdbasource.com> Y Subject: Re: OT but worth posting, virus warning: "FWD: Look at this pack	from Microsoft. 2 Message-ID: <3E5F67AE.1BDE386B@firstdbasource.com>   Kesav Tadimeti wrote:  > G > I had the honour of receving his mail as well. Norton AntiVirus on My 0 > Win2000 m/c detected it and deleted the virus. > Thanks Shane,  > keshav > 9 > +-----------------------------------------------------+  >         KEANE INDIA LIMITED  >         E9 - E12, SDF  >         NEPZ, NOIDA 201 305  >         U.P, INDIA > / >         e-mail: kesav_tadimeti@keaneindia.com % >         phone: +91-120-2568210(371) - >       Men are from MACs, Women are from VMS 9 > +-----------------------------------------------------+   H  I generally at least review the subject in VMS Mail prior to pulling itB with my POP client.  Although I am quite certain Norton would have
 caught it.     --   Regards,  6 Michael Austin            OpenVMS User since June 1984   ------------------------------  % Date: Fri, 28 Feb 2003 10:16:58 -0500 * From: "Syltrem" <syltremzulu@videotron.ca>< Subject: Raise CPUMAXIMUM (CPU time limit) for a running job2 Message-ID: <a4L7a.719$945.2541@tor-nn1.netcom.ca>   Hi  I Would anyone know if there is a way to raise (or void) the CPU time limit  set for a running process?  L The job was submitted on a batch queue that specifies /CPUMAXIMUM = 7 hours,L but the user called in this morning and said it's a special job thay may run< for 10-12 hours. It's been running for 24 hours elapsed now.  + Can this quota be changed for running jobs?    Thanks!    --   OpenVMS 7.2-1, Oracle 8.1.6.0    Syltrem I http://pages.infinit.net/syltrem (OpenVMS related web site - en franais) 8 To reply to myself directly, remove zulu from my address   ------------------------------  % Date: Fri, 28 Feb 2003 09:33:32 -0600  From: brandon@dalsemi.com @ Subject: Re: Raise CPUMAXIMUM (CPU time limit) for a running job1 Message-ID: <03022809333209@dscis6-0.dalsemi.com>    >>> Hi >>> M >>> Would anyone know if there is a way to raise (or void) the CPU time limit  >>> set for a running process? >>> P >>> The job was submitted on a batch queue that specifies /CPUMAXIMUM = 7 hours,P >>> but the user called in this morning and said it's a special job thay may run@ >>> for 10-12 hours. It's been running for 24 hours elapsed now. >>> / >>> Can this quota be changed for running jobs?  >>>  >>> Thanks!   { You will need to set the queue /CPUMAXIMUM=0 (infinite) to allow the job to run then set it back down, see below (DCL HELP)    I am not sure, however you might be able to set the /CPUMAX=0, execute the job, then set the /CPUMAX=7.  At leaset this works with6 setting the job limit, not sure on the /CPUMAX though.     SET      QUEUE        /CPUMAXIMUM              /CPUMAXIMUM=time  A        Defines the maximum CPU time limit for all jobs in a batch H        execution queue. You can specify time as delta time, 0, INFINITE,=        or NONE. You can specify up to 497 days of delta time.   C        The /CPUMAXIMUM qualifier overrides the time limit specified E        in the user authorization file (UAF) for any user submitting a C        job to the queue. Either the value 0 or the keyword INFINITE C        allows unlimited CPU time. If you specify NONE, the CPU time D        value defaults to the value specified either in the UAF or byH        the SUBMIT command (if included). CPU time values must be greaterD        than or equal to the number specified by the system parameter        PQL_MCPULM.  E        For information on specifying delta time, refer to the OpenVMS F        User's Manual or the online help topic DCL_Tips (subtopic Date_
        Time).      SET      ENTRY        /CPUTIME             /CPUTIME=time   D        Specifies a CPU time limit for the batch job. You can specifyH        the time parameter as delta time, the value 0, INFINITE, or NONE.F        If the queue on which the job executes has a defined CPUMAXIMUMH        value, the smaller of the specified job and queue values is used.G        If the queue on which the job executes does not have a specified D        maximum CPU time limit, the smaller of the SUBMIT command andD        user authorization file (UAF) values is used. If the queue onH        which the job executes does not have a specified maximum CPU timeG        limit and the UAF has a specified CPU time limit of NONE, either H        the value 0 or the keyword INFINITE allows unlimited CPU time. IfF        you specify NONE, the specified queue or UAF value is used. CPUH        time values must be greater than or equal to the number specified*        by the system parameter PQL_MCPULM.  F        For information on specifying time values, refer to the OpenVMSF        User's Manual or the online help topic DCL_Tips (subtopic Date_
        Time).        John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Fri, 28 Feb 2003 09:35:25 -0600  From: brandon@dalsemi.com @ Subject: Re: Raise CPUMAXIMUM (CPU time limit) for a running job1 Message-ID: <03022809352565@dscis6-0.dalsemi.com>   / >>> Can this quota be changed for running jobs?   > Now to answer your questions ...  I do not know, I assume not.   John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Fri, 28 Feb 2003 15:44:18 +0000 + From: John Laird <john@laird-towers.org.uk> @ Subject: Re: Raise CPUMAXIMUM (CPU time limit) for a running job8 Message-ID: <mn0v5vg4p2jfl3jhonh01rli2giv8eu6m2@4ax.com>  H On Fri, 28 Feb 2003 10:16:58 -0500, "Syltrem" <syltremzulu@videotron.ca> wrote:  J >Would anyone know if there is a way to raise (or void) the CPU time limit >set for a running process?   F Sure, but you'd have to be into some kernel-mode programming probably,H deliver a special AST to have the process update its own process controlI block (guessing that's where the values are).  A wild guess would be that I whenever the process cpu-used value is updated at the end of a quantum, a J cpu-maximum value will be correspondingly decremented (perhaps only if notK some special unlimited value) - when this goes negative, bang.  Either that I or it's checked by the scheduler when the process comes up for execution.      	John    ------------------------------  % Date: Fri, 28 Feb 2003 11:01:22 -0500 * From: "Syltrem" <syltremzulu@videotron.ca>@ Subject: Re: Raise CPUMAXIMUM (CPU time limit) for a running job2 Message-ID: <OJL7a.726$945.2556@tor-nn1.netcom.ca>   $ set ent/cpu=0 1000388 E %JBC-E-EXECUTING, job executing on queue prevents requested operation   H Can this be done programmatically? Change settings on a running process?! What system service would I call?    --   OpenVMS 7.2-1, Oracle 8.1.6.0    Syltrem I http://pages.infinit.net/syltrem (OpenVMS related web site - en franais) 8 To reply to myself directly, remove zulu from my address  6 <brandon@dalsemi.com> a crit dans le message de news:& 03022809333209@dscis6-0.dalsemi.com... > >>> Hi > >>> I > >>> Would anyone know if there is a way to raise (or void) the CPU time  limit   > >>> set for a running process? > >>> K > >>> The job was submitted on a batch queue that specifies /CPUMAXIMUM = 7  hours,J > >>> but the user called in this morning and said it's a special job thay may run B > >>> for 10-12 hours. It's been running for 24 hours elapsed now. > >>> 1 > >>> Can this quota be changed for running jobs?  > >>> 
 > >>> Thanks!  > J > You will need to set the queue /CPUMAXIMUM=0 (infinite) to allow the job2 to run then set it back down, see below (DCL HELP) > L > I am not sure, however you might be able to set the /CPUMAX=0, execute the7 job, then set the /CPUMAX=7.  At leaset this works with 8 > setting the job limit, not sure on the /CPUMAX though. >  >  > SET  > 	 >   QUEUE  >  >     /CPUMAXIMUM  >  >           /CPUMAXIMUM=time > C >        Defines the maximum CPU time limit for all jobs in a batch J >        execution queue. You can specify time as delta time, 0, INFINITE,? >        or NONE. You can specify up to 497 days of delta time.  > E >        The /CPUMAXIMUM qualifier overrides the time limit specified G >        in the user authorization file (UAF) for any user submitting a E >        job to the queue. Either the value 0 or the keyword INFINITE E >        allows unlimited CPU time. If you specify NONE, the CPU time F >        value defaults to the value specified either in the UAF or byJ >        the SUBMIT command (if included). CPU time values must be greaterF >        than or equal to the number specified by the system parameter >        PQL_MCPULM. > G >        For information on specifying delta time, refer to the OpenVMS H >        User's Manual or the online help topic DCL_Tips (subtopic Date_ >        Time).  >  >  > SET  > 	 >   ENTRY  >  >     /CPUTIME >  >           /CPUTIME=time  > F >        Specifies a CPU time limit for the batch job. You can specifyJ >        the time parameter as delta time, the value 0, INFINITE, or NONE.H >        If the queue on which the job executes has a defined CPUMAXIMUMJ >        value, the smaller of the specified job and queue values is used.I >        If the queue on which the job executes does not have a specified F >        maximum CPU time limit, the smaller of the SUBMIT command andF >        user authorization file (UAF) values is used. If the queue onJ >        which the job executes does not have a specified maximum CPU timeI >        limit and the UAF has a specified CPU time limit of NONE, either J >        the value 0 or the keyword INFINITE allows unlimited CPU time. IfH >        you specify NONE, the specified queue or UAF value is used. CPUJ >        time values must be greater than or equal to the number specified, >        by the system parameter PQL_MCPULM. > H >        For information on specifying time values, refer to the OpenVMSH >        User's Manual or the online help topic DCL_Tips (subtopic Date_ >        Time).  >  >  >  > John Brandon > VMS Systems Administrator  > Dallas Semiconductor > john.brandon@dalsemi.com > 972.371.4172 wk  > 972.371.4003 fx    ------------------------------  % Date: Fri, 28 Feb 2003 10:16:18 -0600  From: brandon@dalsemi.com @ Subject: Re: Raise CPUMAXIMUM (CPU time limit) for a running job1 Message-ID: <03022810161819@dscis6-0.dalsemi.com>    >>> $ set ent/cpu=0 1000388 I >>> %JBC-E-EXECUTING, job executing on queue prevents requested operation  >>> L >>> Can this be done programmatically? Change settings on a running process?% >>> What system service would I call?   7 You can not modify an executing job, even if suspended.   E %JBC-E-EXECUTING, job executing on queue prevents requested operation   : You will need to change the queue and then submit the job.   If you change the entry value (non-executing) then the value will take the lower of the queue or entry (subsequently the queue).   So that will not work.  F As far as system service routine calls - but as John Laird points out   J >>> Sure, but you'd have to be into some kernel-mode programming probably,L >>> deliver a special AST to have the process update its own process controlM >>> block (guessing that's where the values are).  A wild guess would be that M >>> whenever the process cpu-used value is updated at the end of a quantum, a N >>> cpu-maximum value will be correspondingly decremented (perhaps only if notO >>> some special unlimited value) - when this goes negative, bang.  Either that M >>> or it's checked by the scheduler when the process comes up for execution.   / I have not done this ... nor would I want to...   L You might have to delete and re-submit the job under the correct settings...     John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Fri, 28 Feb 2003 10:37:59 -0500  From: G Everhart <ge@gce.com> $ Subject: re CD copy booting off disk$ Message-ID: <3E5F8257.10007@gce.com>  L A copy of a bootable CD onto SCSI disk should boot fine. The driver handlingI between CD and hard drive is almost identical which is why both are dealt D with by the same driver (even at boot time). An ISO filesystem mightE not be understood by the boot filesystem (tho it uses 512 byte blocks E too for reading ISO CDs during normal VMS operation), but then the CD  would not boot either.  G The 512 byte restriction is only there because it takes another command C or two to tell a drive to use that sector size. It looks to me like A that got added to the boot dkdriver about 7.2, though the runtime @ dkdriver has had the logic for years more. Harder to test thingsC in the boot path, so those changes take place more slowly than some  others.   F If you have a CD that refuses to work in 512 byte mode, get IDEZR fromC sigtape or from Hunter's stuff. It will at least let you use the cd B as a data device, with a bit of fiddling to set up (the device youH see becomes a ZR device, and a program does the work to actually twiddleB the underlying CD or other disk; it also will work with disks thatE DKdriver chokes on.) I've used it some (even on some very strange r/w K disks) and had it work ok though a bit slower than a more normal and direct ! disk use due to its data copying.   H While CD error paths differ from scsi disk sometimes, most of the normalL handling of the two as data devices is just the same. Using separate driversN for them would be pretty oddball. On a REAL old box (uVax II maybe) and a realJ old VMS, very very very early dkdriver might have trouble with one and notL the other, but anything more recent should not. More recent dkdrivers should; be better at handling device variety as they get younger...    Glenn Everhart   ------------------------------  % Date: Fri, 28 Feb 2003 14:06:52 +0000 ' From: Andrew Harrison SUNUK Consultancy  Subject: Re: Sun and Java . Message-ID: <3E5F6CFC.2020001@nospamn.sun.com>  , This has been posted already and as you know+ some of the issues were never issues in the * first place and the rest were fixed before Solaris 9 FCS.  1 Hardly a parallel you can draw with the GS160/320 4 which has had these problems all through the product life and the fix is the GS1280.        Rob Young wrote:Z > In article <3E5E4300.8090106@nospamn.sun.com>, Andrew Harrison SUNUK Consultancy writes: >  >>: >>Yes because your own engineers say that there are issues8 >>your own benchmark results illustrate these issues andA >>customers on this newsgroup have benchmarked their applications % >>and discovered them for themselves.  >> >  > & > 	So?  Solaris sucks when using Java: > A > http://www.internalmemos.com/memos/memodetails.php?memo_id=1321  > M > - Large footprint of applications when run on Solaris. A simple application N > ("hello world" type) has a total footprint of 35-40 megs on Solaris 9 (buildP > 48, using Java 1.4 build 82) on both Intel and Sparc machines. Sparc machines,N > by far, have a much higher resident footprint then Intel machines (~30 megs,H > compared to ~11 megs). The same program run on a Windows machine has a: > footprint of ~5 megs, resident footprint being ~3.5 megs > D > 	So maybe you want to run your Java-centric app on Windows, right? > 	Appears that way: > Q > Following is an excerpt from Kevin Tay's e-mail to three Java aliases regarding Q > a customer installation of a third-party product written in Java called Vitria. J > We see typical very large RSS numbers compared to a WinNT implementationC > combined with increased resource usage from Solaris7 to Solaris8: 	 > ------- Q > Customer said they have something like 450+ container servers and 80+ automator P > server for the Vitria system. So the estimation for the hardware RAM is aroundD > 9GB for USII machine and 14-15GB for the USIII machine. Questions: > 2 > 1. Why is Sun systems using so much more memory?N > 2. Why is the UltraSPARC III/Solaris 8 system using a lot more memory than aF > UltraSPARC II/Solaris 7 system (with every other thing being equal)?J > 3. How can I reduce the memory utilization of the UltraSPARC III system? >  > ---  > D > 	Maybe the Java team gets Java healthy under Solaris 8 and SolarisI > 	9.  Or maybe the dirty little secret is Windows NT / 2000 is a better  C > 	OS in comparison?  Intel better than Sparc?  Well, that's no big 
 > 	secret. >  > 	Actually: > P > Imagine what happens if our current implementation of Java were ubiquitous andM > all 150 users on a SunRay server were running one and only one Java program M > equivalent to Component Manager above. The twenty-four gigabytes of RAM the K > server would have to supply exclusively to these users is well beyond the J > typical configuration. RAM is cheap but performance is what we sell, allO > customers on that SunRay server would see significant performance degradation K > even with the maximum amount of RAM installed as all other processes were  > forced to reside on swap.  > O > The resident set size required by the JRE makes it impractical to run Java in J > an initial Solaris install environment. It is impractical to run it as aL > non-terminating daemon. A Java daemon could be started from inetd run longN > enough to do its job and then quit but the rpc protocol required to pass theG > socket port to the daemon is very complex and not Java-friendly. Java J > applications cannot be executed at boot time since the loading of the VMQ > introduces an unacceptable performance degradation. If the Java runtime were as Q > small as that of Python, it is likely that the Java daemon would become popular K > and could provide basic services to applications written in any number of  > languages. >  > D > 	Seems from the tone of that article, a much better solution would@ > 	be to ditch Java and start developing around Python?  Is thatC > 	too much of a stretch or too late to ditch Java?  Very positive   > 	article on Python there!  > 	 > 				Rob  >    ------------------------------  % Date: Fri, 28 Feb 2003 11:29:42 -0500 ! From: Jim Agnew <jpagnew@vcu.edu> 4 Subject: Re: Suspected compromise of my home VMS box' Message-ID: <3E5F8E76.DEE1A8ED@vcu.edu>   E i remember one person saying "they needed a new keyboard" when it was D painfully obvious that coke was responsible.... ;-D  I let them save face..  (sorry guys)   jim    Paul Sture wrote:  > e > In article <yBzw$srjV8EV@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes: Y > > In article <K3KriRP$4iAn@elias.decus.ch>, p_sture@elias.decus.ch (Paul Sture) writes:  > > @ > >> My console keyboard had some sticking keys. I could see the? > >> wrong passwords in the audit log. Small wonder I triggered ' > >> intrusion detection there as well.  > > K > > This problem was fully explained on a US TV documentary* about 15 years  > > ago. > > 
 > > ------/ > > *Saturday Night Live, "The Pepsi Syndrome".  > D > I had a similar problem with a telephone a few years ago, spillingI > Pepsi onto it. It took well over a week for the cradle to become firmly ? > stuck down when the receiver was lifted. Stripping the cradle + > showed a small deposit of Pepsi like goo.  >  > -- > Paul Sture   --  F "4,000 years ago I made a mistake."  Elrond Half-Elven, in "Fellowship of the Ring"   ------------------------------  % Date: Fri, 28 Feb 2003 12:15:14 +0100 8 From: "Tomasz Dryjanski" <tdryjanski.nospam@hotmail.com>$ Subject: SW500 - a hardware question. Message-ID: <b3ngc3$2tb9$1@news2.ipartners.pl>   Hi all, J we have got here an old SW500 with BA-356 ("light-grey") shelves. Normally< they are equipped with 150W (grey and green) power supplies.H Now we want to implement redundant power supply, but there are only 180WJ power supplies available in sale. The salesman claims that we may use 180WL supplies in our shelves, but do you know if we may also mix two power supply types in a single shelf?  " Thank you in advance for any help.   T. D.    ------------------------------  % Date: Fri, 28 Feb 2003 07:08:14 -0600  From: brandon@dalsemi.com ( Subject: Re: SW500 - a hardware question1 Message-ID: <03022807081466@dscis7-0.dalsemi.com>    >>> Hi all, N >>> we have got here an old SW500 with BA-356 ("light-grey") shelves. Normally@ >>> they are equipped with 150W (grey and green) power supplies.L >>> Now we want to implement redundant power supply, but there are only 180WN >>> power supplies available in sale. The salesman claims that we may use 180WP >>> supplies in our shelves, but do you know if we may also mix two power supply >>> types in a single shelf? >>> & >>> Thank you in advance for any help. >>> 	 >>> T. D.   M I have used 180W power supplies in the old narrow shelf without any problems.   J I did not mix the 150W with the 180W though so I do not know the answer toN that.  I would think that it would not matter.  However, just in case, why not9 pair the 150W and then the 180W instead of mixing them?     7 I assume you only have the narrow 4-GB drives, correct?        John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Fri, 28 Feb 2003 16:01:34 +0100 8 From: "Tomasz Dryjanski" <tdryjanski.nospam@hotmail.com>( Subject: Re: SW500 - a hardware question- Message-ID: <b3ntke$57j$1@news2.ipartners.pl>   E > I have used 180W power supplies in the old narrow shelf without any 	 problems.  > L > I did not mix the 150W with the 180W though so I do not know the answer toL > that.  I would think that it would not matter.  However, just in case, why not 9 > pair the 150W and then the 180W instead of mixing them?   H Well, this is not a big problem - it is just spending money for one more
 power supply.   9 > I assume you only have the narrow 4-GB drives, correct?   D We are using N/W drives varying from 2GB to 18GB. Certainly they run
 as narrow.   T. D.    ------------------------------  % Date: Fri, 28 Feb 2003 09:13:12 -0600  From: brandon@dalsemi.com ( Subject: Re: SW500 - a hardware question1 Message-ID: <03022809131224@dscis7-0.dalsemi.com>   G > > I have used 180W power supplies in the old narrow shelf without any  > problems.  > > N > > I did not mix the 150W with the 180W though so I do not know the answer toN > > that.  I would think that it would not matter.  However, just in case, why > not ; > > pair the 150W and then the 180W instead of mixing them?  > J > Well, this is not a big problem - it is just spending money for one more > power supply.   B What I mean by that is, use 150W or 180W per shelf but not both.    ; > > I assume you only have the narrow 4-GB drives, correct?  > F > We are using N/W drives varying from 2GB to 18GB. Certainly they run > as narrow. >  > T. D.   O Not good, I should have specified the speed rating.  I am not sure of the speed M of the 18GB, however, you probably need to be at 180W for anything about 7K.  J Do not mix the 150W and 180W - a 150W is not capable of maintaining a full shelf of high speed drives.        John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Fri, 28 Feb 2003 16:58:06 +0100 8 From: "Tomasz Dryjanski" <tdryjanski.nospam@hotmail.com>( Subject: Re: SW500 - a hardware question- Message-ID: <b3o0uh$7k3$1@news2.ipartners.pl>   K > Not good, I should have specified the speed rating.  I am not sure of the  speed J > of the 18GB, however, you probably need to be at 180W for anything about 7K.     Well, it's good to know. Thanks.H So far we haven't filled any shelf with 7K only, and we will avoid this.   T. D.    ------------------------------  % Date: Fri, 28 Feb 2003 11:08:10 -0500 ! From: VAXVMS <bounce@notmail.com> ( Subject: RE: SW500 - a hardware questionK Message-ID: <BA52530E3149734A9BAABDBBFA808E4903027BCE@rlghncst964.usps.gov>   G > > I have used 180W power supplies in the old narrow shelf without any  > problems.  > > K > > I did not mix the 150W with the 180W though so I do not know the answer  toJ > > that.  I would think that it would not matter.  However, just in case, why  > not ; > > pair the 150W and then the 180W instead of mixing them?  > J > Well, this is not a big problem - it is just spending money for one more > power supply.   @ What I mean by that is, use 150W or 180W per shelf but not both.  ; > > I assume you only have the narrow 4-GB drives, correct?  > F > We are using N/W drives varying from 2GB to 18GB. Certainly they run > as narrow. >  > T. D.   I Not good, I should have specified the speed rating.  I am not sure of the  speed L of the 18GB, however, you probably need to be at 180W for anything about 7K.J Do not mix the 150W and 180W - a 150W is not capable of maintaining a full shelf of high speed drives.        John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx   F If I remember correctly, there are also two variants of fans for these units:- A regular and an extra-strength, so to speak.   L I wouldn't load up a shelf with faster drives unless I made sure it had the  extra-strength fans.   ========================  William W. Webb - EMS Operations) OpenVMS Systems Support - USPS DSSC Annex , 4730 Hargrove Road, Raleigh, NC 27616-2874  > 919.325.7500 x4186  <FirstInitialLastNameAtEmailDotUSPSDotGov>   ------------------------------  % Date: Fri, 28 Feb 2003 11:50:59 +0000 - From: Roy Omond <Roy.Omond@BlueBubble.UK.Com> ( Subject: Re: Upcoming VMS improvements ?/ Message-ID: <v5uj9drbq8lha7@corp.supernews.com>    Larry Kilgallen wrote:  D > In article <3E5CE9D5.80601@MMaz.com>, "Barry Treahy, Jr."  writes: >  > >Larry Kilgallen wrote:  > >  > >  > [... snip ...] > >>C > >>What is the VAX emulation configuration that produces 60 VUPs ?  > >> > >> > >  > >Dual AMD 2600+ MP...  > ( > Does it emulate a dual processor VAX ? > 0 > >Talk to SRI, they have been doing some rather > >impressive marvels... > G > My question is only based on curiosity, and their "prices are secret" H > policy, imposed even on resellers, tells me they are not interested in  > satisfying the merely curious.  B Larry, that's been my experience also here in the UK.  I *did* getA sort of a price quotation, but I was so shocked at how big it was : (even for a minimal system, which according to the quoter,  @ "The product price range is from about n.nK (absolute minimum no@ one has done yet) to nnK depending on how hard it is to complete, and how many VUPs the client finally needs."  C I gave up interest).  I feel obliged to substitute the real numbers 9 with "n" just in case ...  these prices are in UK pounds.   	 Roy Omond  Blue Bubble Ltd.   ------------------------------  # Date: Fri, 28 Feb 2003 14:43:15 GMT # From: "John Smith" <a@nonymous.com> ( Subject: Re: Upcoming VMS improvements ?G Message-ID: <7AK7a.11657$em1.8315@news04.bloor.is.net.cable.rogers.com>   = "JF Mezei" <jfmezei.spamnot@vl.videotron.ca> wrote in message ) news:3E55AFE2.4144551C@vl.videotron.ca... < > Has Digital/compaq/HP announced any plans to improve VMS ?     My vote:  , 1) Effective advertising & marketing for VMS> 2) Control of the advertising budget 100% within the VMS groupE 3) Bonuses tied to increased marketshare and profits for the VMS team  (secretaries up to Marcello) 4) a new BOD 5) EV8   ------------------------------  # Date: Fri, 28 Feb 2003 15:01:27 GMT # From: "John Smith" <a@nonymous.com> ( Subject: Re: Upcoming VMS improvements ?G Message-ID: <bRK7a.11947$em1.9612@news04.bloor.is.net.cable.rogers.com>   D "Fred Kleinsorge" <kleinsorge@star-dot-zko-dot-dec-dot-com> wrote in0 message news:3e5e50f2$1_2@hpb10302.boi.hp.com...@ > The fact it's on the roadmap doesn't indicate that development starts today@ > and ends at that date.  That is the delivery date for when the driver will E > be completed.  As 10GB is verrry expensive backbone hardware today,  it is D > lower on the priority list than other things, and hence the driver	 work, and 7 > decision on what to support, won't start for a while.     ' Just being devil's advocate (as usual):   D Only those companies/government agencies with money to burn would be@ early adopters of 10+ Gigabit Ethernet. Wouldn't that make theseD organizations the ideal customers for clusters of GS1280's and theirF associated high profit margins. Seems to me that if you got the driverA done for these today HP might be better off financially. Not many C lower margin IA-64 systems with VMS on them will ship initially, so 1 what if they are a few months later out the door?    ------------------------------  % Date: Fri, 28 Feb 2003 10:25:41 -0500 - From: "Peter Weaver" <peter.weaver@stelco.ca> ( Subject: Re: Upcoming VMS improvements ?6 Message-ID: <b3nv1s$1o5mtf$1@ID-141708.news.dfncis.de>   Alan E. Feldman wrote: >... > I want to do:  > 
 > Gold-PF3 > Forward Find: STRING > G > and have it find all occurrences of regardless of the cases of any of D > its letters. I want it to Find the strings string, sTRING, String,D > STRING, etc. But the help you refer me to tells me that I can't do > that. I have to instead do >...  J Here you go Alan, Copy the procedure below into your paste buffer, go to a VMS window, type the commands  $ SET TERM/HOST  $ EDIT/EDT ALAN.COM L then paste your buffer into the window, exit the editor and type @ALAN. EditI a file with this newly created TPU section and try searching for "AS" and J "As" and "as" - you should find all three lines each time. (I know this isI not all that you are after, but this might make you life a bit easier the . few times where you need to get into EVE. :) )  K (watch the wrapping, I have wrapping set for 76 characters and each line is L less than 76 so it should not be a problem, but I am using OE to post this.)   $!+ $ create sys$login:new_find_target_case.tpu E procedure eve$set_find_target_case      ! Set target's case exactness      (how_exact)   > ! EVE's FIND is case-sensitive unless (1) sensitivity has been? ! set to noexact, and (2) the target is all lowercase, in which & ! case the search is case-insensitive.  G ! Modified to remove case (2) 28-FEB-2003 so Alan will be a bit happier   C local   lowercase_target,       ! Lowercase version of eve$x_target          the_target;   + if get_info (eve$x_target, "type") = STRING  then%     lowercase_target := eve$x_target;  else-     lowercase_target := eve$x_target_pattern;  endif; the_target := lowercase_target;   & change_case (lowercase_target, LOWER); how_exact := eve$x_find_exact;$ if not eve$get_find_case_sensitivity then7 !    if lowercase_target = the_target ! Remove case (2) 7 !    then                             ! Remove case (2) )         how_exact := eve$x_find_no_exact; 7 !    endif;                           ! Remove case (2)  endif;   return (TRUE);  ( endprocedure; ! eve$set_find_target_case $!9 $ edit/tpu/nodisplay sys$login:new_find_target_case.tpu -       /command=sys$input  eve_extend_tpu ("*") ; eve_set_find_case_noexact ; , eve_save_extended_tpu ("sys$login:new_eve"); $! $ create test_new_search.txt Upper AS Lower as Mixed As $!( $ assign sys$command sys$input/user_mode8 $ edit/tpu/section=sys$login:new_eve test_new_search.txt $! $ write sys$output -9   " Modify your EDIT symbol to point to the section file"  $ write sys$output -7   " sys$login:new_eve or copy the new section file over  $ write sys$output -   " your old section file."  $!     -- Peter WeaverD Opinions are my own, and do not reflect the opinions of my employer,A nor the company that it sub-contracts to, nor the company that it  sub-contracts to.    ------------------------------  % Date: Fri, 28 Feb 2003 15:35:33 -0000 * From: "Richard Brodie" <R.Brodie@rl.ac.uk>( Subject: Re: Upcoming VMS improvements ?+ Message-ID: <b3nvk7$vpi@newton.cc.rl.ac.uk>   . "John Smith" <a@nonymous.com> wrote in messageA news:bRK7a.11947$em1.9612@news04.bloor.is.net.cable.rogers.com...   G > Seems to me that if you got the driver  done for these today HP might  > be better off financially.  G This discussion seems to have missed an obvious point or two: as far as D I know, nobody actually sells 10-Gigabit NICs yet. And when they do,8 its likely that PCI isn't going to be the bus of choice.  I So RAIN across multiple PCIs looks like a no-brainer for Alpha, at least.    ------------------------------  % Date: Fri, 28 Feb 2003 11:58:04 -0500 . From: "Leo Demers" <leo_dot_demers@HP_dot_com>( Subject: Re: Upcoming VMS improvements ?, Message-ID: <3e5f96f9$1@usenet01.boi.hp.com>  I Ok. Since I'm the Product Manager for the LAN adaptors perhaps I can shed D some light on the roadmap process by using 10Gig as an example here.  I First and foremost the roadmap slide has a little arrow in the upper left I hand corner that shows you that OpenVMS will continue beyond 2005 this is H just what we can fit on today's roadmap. If you saved an OpenVMS roadmapH from back in 2000 it only showed releases up until 2003. Since we're allL working on getting multiple VMS releases ready right now historical evidenceJ shows that it OpenVMS releases continue to happen beyond what is currentlyJ visible in the roadmap window.  Practically  we can only plan out with anyK sort of accuracy a release schedule for a couple of years.  Back in 2000 we L didn't have a little thing like the Itanimum releases on our roadmap, so you; can see how things change as we try to roadmap farther out.   L  For 10 Gigabit Ethernet the reason it is placed on the roadmap in 2005 is aJ factor of the industry pricing, engineering/testing resources lining up onK an OpenVMS release. In this case we have looked at the industry and believe H that 10Gig is something OpenVMS customers will want as a NIC and ClusterG Interconnect. We also know that the cost trends in the industry and are J watching them. Right now the cost per port for 10Gig is something close toJ $2K per NIC and nearly twice that per port on a switch so cost is a factorJ here as well as selecting a vendor who's cards will perform and that'll beJ around in a few years.  It might be cool to sell one $2K NIC to a customerH who's looking to push 4-5Gigs but that's not a very sustainable businessL model when you factor in our engineering and testing efforts. Alphas tend toJ push the envelop of the NICs designs more the standard Intel32 boxes do so@ it takes time to shakeout the driver and card and test it in ourC environment.   Then the price will come down from a premium cost to ) commodity pricing  in a 1-2 year window.. L  Factoring all of that in with what we currently believe to be a VMS releaseK schedule puts it on the roadmap and that's just one little box for a fairly + straightforward (engineering wise) project.     - Leo --
 Leo Demers  OpenVMS Security Product Manager Leo_dot_Demers_at_HP_dot_COM. "John Smith" <a@nonymous.com> wrote in messageA news:bRK7a.11947$em1.9612@news04.bloor.is.net.cable.rogers.com...  > F > "Fred Kleinsorge" <kleinsorge@star-dot-zko-dot-dec-dot-com> wrote in2 > message news:3e5e50f2$1_2@hpb10302.boi.hp.com...B > > The fact it's on the roadmap doesn't indicate that development > starts todayB > > and ends at that date.  That is the delivery date for when the
 > driver will G > > be completed.  As 10GB is verrry expensive backbone hardware today,  > it is F > > lower on the priority list than other things, and hence the driver > work, and 9 > > decision on what to support, won't start for a while.  >  > ) > Just being devil's advocate (as usual):  > F > Only those companies/government agencies with money to burn would beB > early adopters of 10+ Gigabit Ethernet. Wouldn't that make theseF > organizations the ideal customers for clusters of GS1280's and theirH > associated high profit margins. Seems to me that if you got the driverC > done for these today HP might be better off financially. Not many E > lower margin IA-64 systems with VMS on them will ship initially, so 3 > what if they are a few months later out the door?  >  >    ------------------------------    Date: 28 Feb 2003 09:21:50 -0800. From: spamsink2001@yahoo.com (Alan E. Feldman)( Subject: Re: Upcoming VMS improvements ?< Message-ID: <b096a4ee.0302280921.11b9099@posting.google.com>  g JF Mezei <jfmezei.spamnot@vl.videotron.ca> wrote in message news:<3E5E8871.F854FFC5@vl.videotron.ca>...  > "Alan E. Feldman" wrote:D > > I don't have your editor. And I don't want to waste time and getG > > aggravated trying to write it when I already have something I like.  > ) > You're either with us (TPU) or against. ' > EDT will be disarmed, no matter what. N > We are, at the moment, amassing troups of TPU experts around your IP addressM > armed with all their customized TPU tools and programs, ready to invade and O > impose a TPU regime to finally get rid of that terrorist EDT regime which has P > continued to produce/hide commands of mass text manipulation which is a dangerI > to the world and which has ties with the worst bunch of terrorists, the  > Al-qUnix faction.   E It started out funny, but quickly became tiresome. Obviously, JF, you F have dreams of grandeur of becoming another BG, shoving bloatware down people's throats. :-)    [...] N > And as a further measure to ensure homeland security against EDT terrorists,P > HP has announced it has licenced software from Microsoft that will be embeddedI > in all VMS systems to automatically notify Homeland Security of any/all L > attempts to start EDT on any machine, with all the personal details of the9 > user being transmitted automatically over the internet.   B Ah, more evidence confirming my theory that you are a BG wannabee.  W > Resistance is futile... (oh, sorry, that is from another non "editor-jihad" universe)  >  > :-) :-) :-) :-)   F Yeah, yeah, yeah. Use whatever editor you like, I don't care. I'll useC what I like. And if you don't like it, you can write a TPU program!    Disclaimer: JMNSHO Alan E. Feldman    ------------------------------    Date: 28 Feb 2003 12:12:05 -0600+ From: young_r@encompasserve.org (Rob Young) ( Subject: Re: Upcoming VMS improvements ?3 Message-ID: <FJrsqC7bPFZ+@eisner.encompasserve.org>   ] In article <3e5f96f9$1@usenet01.boi.hp.com>, "Leo Demers" <leo_dot_demers@HP_dot_com> writes:   = > Right now the cost per port for 10Gig is something close to L > $2K per NIC and nearly twice that per port on a switch so cost is a factorL > here as well as selecting a vendor who's cards will perform and that'll be > around in a few years.  F 	Do you have a reference for those 10 Gig port and switch prices?  TheA 	reason I ask, this author must have badly butchered his research  	on the subject:  [ http://www.computerworld.com/networkingtopics/networking/lanwan/story/0,10801,76979,00.html    N "Although most 10 Gigabit Ethernet offerings fall into the $50,000 [per port] N range, Enterasys Networks' recent rollout, priced under $30,000, could be the & start of a downward trend in pricing."  
 	Likewise:  [ http://www.computerworld.com/networkingtopics/networking/lanwan/story/0,10801,76977,00.html   K "But even though the Institute of Electrical and Electronics Engineers Inc. M approved the final 10 Gigabit Ethernet standard last June, the price of entry J -- at an average of $50,000 per port -- has kept most organizations on the sidelines."    [snip]  J "By 2006, the cost [for 10 GBit enet] will be $5,000 to $6,000 per port," L predicts Chandra Kopparapu, director of product marketing at San Jose-based $ switch vendor Foundry Networks Inc."   				Rob    ------------------------------  # Date: Fri, 28 Feb 2003 14:00:47 GMT 2 From: "Robert Boers" <robert.boers@softresint.com> Subject: Re: VAX Emulators* Message-ID: <3e5f6b76$1@news.deckpoint.ch>  L For Windows 2000/XP systems, performance varies a bit between P4, Xeon 4 andF AMD systems, which have different fixed point, floating point or cache characteristics.  J For example, a dual AMD 2200MP+ system, 1 GB, Windows XP, produces 64 VUPsF when running CHARON-VAX/XL+ (emulatiing a 512 MB VAX 3100-98). This is comparable with a VAX 7810.   F On a DS25 (OpenVMS 7.3) running CHARON-VAX/AXP (emulating a 512 MB VAX" 3100-98) we measure about 30 VUPs.   Regards, Robert Boers  . "Tom Linden" <tom@kednos.com> wrote in message3 news:CIEJLCMNHNNDLLOOGNJICEEHGLAA.tom@kednos.com... = > Does any have data on comparison of VAX emulator running on ? > Pentium to Alpha?  In other words, what must the frequency of ; > an Alpha be for it offer comparable performace to, say, a = > 2.7GHz Pentium (with scsi and all that to offer a more fair 
 > comparison)  > --- ( > Outgoing mail is certified Virus Free.< > Checked by AVG anti-virus system (http://www.grisoft.com).B > Version: 6.0.449 / Virus Database: 251 - Release Date: 1/27/2003 >    ------------------------------  % Date: Fri, 28 Feb 2003 06:15:51 -0800 # From: "Tom Linden" <tom@kednos.com>  Subject: RE: VAX Emulators9 Message-ID: <CIEJLCMNHNNDLLOOGNJIKEGAGLAA.tom@kednos.com>   K Thanks Robert.  I am aware of the shortcommings of benchmarking, but what I  was , looking for was something like the following  C 100Mhz VAX = x MHz Alpha = VAX EMUL on y Mhz Intel or AMD something  x=?  y=?   8 Test should probably include VAX Floating types as well.   Do you have any published data?    Tom    >-----Original Message----- 8 >From: Robert Boers [mailto:robert.boers@softresint.com]( >Sent: Friday, February 28, 2003 6:01 AM >To: Info-VAX@Mvb.Saic.Com >Subject: Re: VAX Emulators  >  > B >For Windows 2000/XP systems, performance varies a bit between P4, >Xeon 4 and G >AMD systems, which have different fixed point, floating point or cache  >characteristics.  > K >For example, a dual AMD 2200MP+ system, 1 GB, Windows XP, produces 64 VUPs G >when running CHARON-VAX/XL+ (emulatiing a 512 MB VAX 3100-98). This is  >comparable with a VAX 7810. > G >On a DS25 (OpenVMS 7.3) running CHARON-VAX/AXP (emulating a 512 MB VAX # >3100-98) we measure about 30 VUPs.  > 	 >Regards, 
 >Robert Boers  > / >"Tom Linden" <tom@kednos.com> wrote in message 4 >news:CIEJLCMNHNNDLLOOGNJICEEHGLAA.tom@kednos.com...> >> Does any have data on comparison of VAX emulator running on@ >> Pentium to Alpha?  In other words, what must the frequency of< >> an Alpha be for it offer comparable performace to, say, a> >> 2.7GHz Pentium (with scsi and all that to offer a more fair >> comparison) >> ---) >> Outgoing mail is certified Virus Free. = >> Checked by AVG anti-virus system (http://www.grisoft.com). C >> Version: 6.0.449 / Virus Database: 251 - Release Date: 1/27/2003  >> >  >  >---' >Incoming mail is certified Virus Free. ; >Checked by AVG anti-virus system (http://www.grisoft.com). A >Version: 6.0.449 / Virus Database: 251 - Release Date: 1/27/2003  >  --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.449 / Virus Database: 251 - Release Date: 1/27/2003   ------------------------------    Date: 28 Feb 2003 08:40:32 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: VAX Emulators3 Message-ID: <o4F7GJm275ou@eisner.encompasserve.org>   _ In article <3e5f6b76$1@news.deckpoint.ch>, "Robert Boers" <robert.boers@softresint.com> writes: N > For Windows 2000/XP systems, performance varies a bit between P4, Xeon 4 andH > AMD systems, which have different fixed point, floating point or cache > characteristics. > L > For example, a dual AMD 2200MP+ system, 1 GB, Windows XP, produces 64 VUPsH > when running CHARON-VAX/XL+ (emulatiing a 512 MB VAX 3100-98). This is > comparable with a VAX 7810.  > H > On a DS25 (OpenVMS 7.3) running CHARON-VAX/AXP (emulating a 512 MB VAX$ > 3100-98) we measure about 30 VUPs.  ; So what is the contribution of these being dual processors, D if the result is comparable to a 7810 rather than a 7820 (or 6320) ?   ------------------------------  # Date: Fri, 28 Feb 2003 15:07:10 GMT 2 From: "Robert Boers" <robert.boers@softresint.com> Subject: Re: VAX Emulators* Message-ID: <3e5f7b1b$1@news.deckpoint.ch>  E Dual host processors are required (unless you do very little I/O) for I balanced production systems. The emulators are multithreaded, and the VAX D CPU emulator component likes to occupy a whole host CPU. The current/ emulator models are all single CPU VAX systems.   H With a second host CPU, I/O can be very fast. With SCSI-160 disks we getL VAX/VMS I/O speeds of 20 - 30 MB/sec. In a test of CHARON-VAX/AXP on a largeI Alphaserver with an EMC backend we hit 50 MB/sec with files of 1 MB size, K but that is exceptional. Even on a good laptop the I/O exceeds the hardware F VAX I/O by a long way (but there is 20 years of technology in between)  L We don't guarantee performance, but it works on a Windows laptop OK. Compile9 speed on a 2.5 GHz laptop is about the same as on a 7610.    Regards, Robert     : "Larry Kilgallen" <Kilgallen@SpamCop.net> wrote in message- news:o4F7GJm275ou@eisner.encompasserve.org... ; > In article <3e5f6b76$1@news.deckpoint.ch>, "Robert Boers" % <robert.boers@softresint.com> writes: L > > For Windows 2000/XP systems, performance varies a bit between P4, Xeon 4 and J > > AMD systems, which have different fixed point, floating point or cache > > characteristics. > > I > > For example, a dual AMD 2200MP+ system, 1 GB, Windows XP, produces 64  VUPsJ > > when running CHARON-VAX/XL+ (emulatiing a 512 MB VAX 3100-98). This is > > comparable with a VAX 7810.  > > J > > On a DS25 (OpenVMS 7.3) running CHARON-VAX/AXP (emulating a 512 MB VAX& > > 3100-98) we measure about 30 VUPs. > = > So what is the contribution of these being dual processors, F > if the result is comparable to a 7810 rather than a 7820 (or 6320) ?   ------------------------------  % Date: Fri, 28 Feb 2003 16:30:29 +0000 ' From: Andrew Harrison SUNUK Consultancy ! Subject: Re: VMS Backup solutions . Message-ID: <3E5F8EA5.8080306@nospamn.sun.com>   Bob Koehler wrote:O > In article <03022615422724@dscis6-0.dalsemi.com>, brandon@dalsemi.com writes:  >  >>Our VMS solution is to use BACKUP with DCL command files.  Automated using ISAM files and scripts with logging, e-mail, reporting,| >>WWW, paging, etc.  All in-house created and maintained.  I have worked with SLS in the past and incorporated many of those- >>concepts into the in-house backup solution.  >>q >>We keep our tapes at a 95% to 98% capacity so we make efficient use of our tapes, no multi-volume tapes either.  >> >>Unlike the NT or UNIX group, the VMS group has maintained a stable backup solution incorporated with DR practices and policies.  >  > J >    So they're trying to solve a problem that doesn't exist.  NT and UNIXI >    backup utilties are crap, or non-existant.  So they're assuming they / >    need to include VMS in the solution.  PHM.  >    Humm  A The Tape SAN that we have implimented at the retail customer I am > engaged with at the moment supports around 120 Sun servers 20+ SF6800 and 5 SF15000.   ; We have two STK Silos reach with 20 9740 FC drives. Current 8 storage footprint is ~60 TB with ~5 x that in the silos.  , The largest backup job runs in at >100 MB/s.  ( All managed using UNIX backup utilities.   Regards  Andrew Harrison    ------------------------------  % Date: Fri, 28 Feb 2003 01:42:32 -0500 * From: "Bill Todd" <billtodd@metrocast.net>$ Subject: Re: Volume set across RAID52 Message-ID: <MuGdnfa6rfpAmcKjXTWcog@metrocast.net>  < "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message  news:3E5ED96E.A6C8873@fsi.net... > Bill Todd wrote: > > @ > > "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message% > > news:3E5D870E.50790DA3@fsi.net...  > > > Bill Todd wrote: > > > > D > > > > "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message) > > > > news:3E5C272E.EFFEA73E@fsi.net...  > > > > > jlsue wrote: > > > >  > > > > ...  > > > > H > > > > > > It's just simply amazing the simplification that the EVA can bring.	 > > > > > L > > > > > I tend to view that rather like the choice between burning garbage orF > > > > > burying it in a landfill. The bad stuff has to go SOMEplace.	 > > > > > I > > > > > Rather like outsourcing: that heat may be taken off of you, but  how  > > muchL > > > > > do you trust those upon whom the burden now falls? ...and how will you B > > > > > respond should that trust be given and then be breached? > > > > J > > > > Don't you think that you really ought to learn something about EVA
 > > before2 > > > > presuming to voice such opinions about it? > > > D > > > My point was that the management effort *HAS* to go someplace. > > L > > The only point you're making is that you not only know nothing about EVA but K > > are unwilling to step back and take a look at it when your ignorance is  > > pointed out. > I > Well, I've learned not to contradict when your mind is made up; so, ...   H But even now you obviously haven't learned that when you don't know shitD about something and someone calls you on it, persisting in both yourI ignorance and your uninformed viewpoint just makes you look both ignorant  and incompetent.   >  > >  You canK > > > either manage the storage farm, or manage the stuff that manages your J > > > storage farm for you, but either way, you're gonna have your hand in the 
 > > > pie. > > K > > Wrongo, Dave.  With EVA, you *don't* have to have your hand in the pie.  ItK > > really does take care of the problem:  about all you need to do is feed  it3 > > more disks when space runs low or a disk fails.  > I > Hhmmm... Sure sounds like there's still some management effort there to  > me  H While (as I already noted) you still need to feed it disks when it needsH them, the statement you seemed to have trouble accepting above was "It'sE just simply amazing the simplification that the EVA can bring."  Your I response suggested that one way or another you'd still need to do just as  much work.  You were wrong.   (  ... unless the thing configures itself,  J To a reasonable degree, it does.  Though it obviously can't read your mindK about what LUNs you may want and what limits you might wish to place on the / amount of disk space will be dedicated to each.    > optimizes itself,    That it does quite well.    splits 4 > mirror-sets and/or creates snap-shots psychically,  F Ask for a snapshot and you get it - without the need for an additionalD mirror set (i.e., additional physical space requirements are minimalC compared with an additional mirror set unless you're updating large F fractions of the data set during the period while the snapshot remains active).    interfaces withE > DECevent and/or Analyze so we can get automated status reports when # > arrive at work every morning, ...   D Haven't a clue about that one:  it probably has reasonable reportingK capabilities, and the fact that operation is mostly automated should reduce I the need for much in the way of status reports save for failed components  and space-low conditions.    > J > ...but, your mind is made up, and you're always right, so guess I'm down" > the for the count as usual, huh?  H Unless you pick yourself up and actually deign to learn something rather! than just nurse your bruised ego.    - bill   ------------------------------  % Date: Fri, 28 Feb 2003 06:38:48 -0500   From: John Santos <JOHN@egh.com>$ Subject: Re: Volume set across RAID56 Message-ID: <1030228061707.26948A-100000@Ives.egh.com>  - On Thu, 27 Feb 2003, David J. Dachtera wrote:  [...]    > E > Not vaulable to me unless there's some SUPPORTED way to add a third H > mirror-set member(-set), split that mirror off and present it as a new> > unit or backup from within an OpenVMS batch job (lights-out, > unattended).  = If I understand EVA correctly, I think it presents a bunch of B virtual disks to the host(s), much as the older RAID-5 controllersA can do.  You get to decide how big each virtual disk is, somewhat ; like partitioning a disk on a PC or in (many, all?) Unices.   B The virtual disks are spread across all spindles, with redundancy,, but you don't have to worry about that part.  B If you have a database disk you want to snapshot, I think what you@ do is take two partitions of the same size and shadow them (with; traditional VMS volume shadowing).  When backup time comes, = quiese the application, close all the files (or have it flush @ all pending I/O), split the shadow set by dismounting one member= (or dismount the shadow set and remount just one member), and = resume the application.  This should only take a few seconds.   ; Then remount the split-off snapshot member, back it up, and  re-add to the shadow set.   < This is just like doing the same thing now, except the units< of the shadow sets aren't real disks anymore, but are chunks of EVA storage.   7 (BTW, 2 extra points about this snap-shot method... The 8 application needs to be able to record its status in its8 journal or log files at the time of the snapshot, if you8 want to be able to recover after restoring the snapshot.9 A backup is useless if you don't know exactly when it was 5 performed!  Second, it is possible to write an app to 9 record its progress to a transaction log, rather than the 8 real database, so that processing can continue (maybe at: a degraded rate) while the flushing/dismounting/remounting: is going on, so the users don't actually see any down time6 at all.  I think most (all?) DBMS systems use these or8 similar methods to make online backups, but you can also roll your own using RMS.)    [...]    --   John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  % Date: Fri, 28 Feb 2003 09:35:44 -0500 & From: jlsue <jlsuexxxz@screaminet.com>$ Subject: Re: Volume set across RAID58 Message-ID: <32su5v05gbelbn4q5tanj5j5ve693ai431@4ax.com>  D On 27 Feb 2003 10:34:14 -0600, young_r@encompasserve.org (Rob Young) wrote:  b >In article <gtas5v81eruoldi6vd0n032h1a0tmnp24h@4ax.com>, jlsue <jlsuexxxz@screaminet.com> writes: >  >>  B >> What it's done for databases is eliminate the need to regularlyI >> re-partition/position data on LUNs to address I/O bottlenecks - a very 0 >> manual, time-consuming, often off-hours task. >>   >  >	EVA is a winner all around.  > H >> Couple this with the CASA solution that allows you to virtualize yourH >> LUNs across controllers, even over IP WAN, and you've got some prettyG >> cool capabilities for high availability solution as well (far better $ >> than the EMA DRM solution, imho). > < >	CASA is kind of an odd bird.  It is neat in some respects,H >	troublesome in others.  RAID1 across datacenters is good, you would be9 >	beyond full merge problem with shadowing.  However, VMS  >	isn't supported yet: >   F True, but it's in the roadmap for "potential" support later this year.F Sorry I didn't state that more clearly.  Anyone who hasn't implemented9 multi-site shadowing yet may be able to plan around this.    > B >	Secondly, others have pointed out (traderags, let me duck) that H >	CASA in largish situations could be a choke point as it is constrained* >	to 80000 IOPS and 800 MB/sec throughput. > < >	This brochure shows on page 7, you may want 2 of them for  >	availability:   B Yup.  There's are trade-offs between performance, reliability, and@ cost.  Always has been, always will be.  HBVS also has overhead,? though I you don't see many numbers that describe the resulting  effect.    > d >http://www.hp.com/products1/storage/pdfs/virtualization_appliances/CASA_Whitepaper_FINAL_110602.pdf > 8 >	Thirdly, CASA is not cheap.  This is my biggest issue. > 
 >	Page 14: > O >"In a CASA SAN with additional switches, up to approximately 20 hosts and four L >midrange storage systems or a single enterprise-class storage system is the$ >recommended maximum configuration."  E Somewhat true.  It is very dependent on your environment - so the EVA C for that matter.  In some environments the EVA vs EMA shows the EVA F more expensive.  In others, with the resulting number of EMAs it takesE to provide the same level of service as one EVA, the EVA is the clear  winner.   E Likewise, if you're going to mirror across long distances, you'll use D something like, for example, a pair of CNT boxes.  These also aren'tB cheap it turns out, and the functionality the allow (DRM) isn't as flexible or as transparent.     E And for many folks, the virtualization - which can provide Continuous B Access to the Storage - is worth for more to the business than the cost of the Appliance.  B So I hesitate to say that one is a clear winner over the other.  I> spend a lot of time these days helping customers analyze theirE specific situation to determine what works best for their businesses.    > @ >	Single EVA?  One EVA per CASA?  Well sure as two EVAs (or moreB >	technically 4 EVA controllers) would be constrained by the CASA ! >	chokepoint.  There is an issue.  > ; >	Divide that out per server and it is a pricey solution.   I >	"Yes, but what price availability?  Can you afford to be without it?"   4 >	Good points, but still a large issue cost.   ($$$)  E Exactly.  The cost/benefit analysis can be very detailed.  More often C than not, you aren't going to replicate *all* your data anyway.  So B the actual number of systems that need to be connected to the CASA> (for remote replication, which is, admittedly, only one of theE applications for virtualization) may be relatively small.  And in the C remote replication biz, the line speeds may be a bigger chokepoint.   F For virtualization within a data center, well it can be a bit tougher.? The value of availability can be worth much, much more than the 1 product.  But there are limitations (always are).    > > >	It is a nice product and I'm sure you are selling it, but toD >	me it would make a whole lot more sense if the out-of-band portion, >	of virtualization was quite a bit cheaper. > ? >	Finally, rock solid new method of Shadow mini-merge (storage  G >	independent, tying into a new filesystem?) would in many ways render   >	CASA irrelevent for VMS. >   % Possibly, we'll have to wait and see.    ------------------------------  % Date: Fri, 28 Feb 2003 09:47:28 -0500 & From: jlsue <jlsuexxxz@screaminet.com>$ Subject: Re: Volume set across RAID58 Message-ID: <uusu5vsenbuat6p932j88qqfhgd2hq14g8@4ax.com>  7 On Thu, 27 Feb 2003 21:30:22 -0600, "David J. Dachtera"  <djesys.nospam@fsi.net> wrote:  
 >jlsue wrote:  >>    > >> Well, sure, but that's why they put the intelligence in theE >> controller.... to save YOU time and energy.  The management of the @ >> storage is now automated, and optimized for performance *and*A >> reliability.  (For example, when you create a VRAID-1 LUN, the A >> controller takes care that LBN 'x' is mirrored on a completely ! >> different drive in the group.)  > D >Not vaulable to me unless there's some SUPPORTED way to add a thirdG >mirror-set member(-set), split that mirror off and present it as a new = >unit or backup from within an OpenVMS batch job (lights-out, 
 >unattended).   @ This is an interesting point of discussion.  People who are veryD comfortable with the older JBOD-like lun presentation will recognize4 this sentiment - and I also fall into this category.  E EVA provides other ways to do the same thing.  It's not *exactly* the E same as cloning under HSJ/D/Z/G controllers, but it provides the same ( functionality from a backup perspective.  F You'd typically use snapshot or snapclone to do this with EVA.  It can2 be automated for "lights-out", unattended backups.   > D >> It appears that you have your doubts about it and don't trust it, > H >No, everytime "they" come up with a new way to make my life "easier", IH >have to spend six or seven man-months trying to find ways to circumventH >it so I can get what I need out of it without sacrificing reliablility.  E Yes, I've found myself in that position a few times as well.  In this @ case, though it doesn't technically work the exact same way, theB overall functionality is there.  The main part missing is that youD can't identify which specific disks have the data (well, actually itE spreads the data all over).  It can be disconcerting at first because  it is significantly different.  C I've run through my own doubting sessions with it before getting to E where I am today on the subject.  It's pretty fascinating technology,  really.    > B >> What it's done for databases is eliminate the need to regularlyI >> re-partition/position data on LUNs to address I/O bottlenecks - a very 0 >> manual, time-consuming, often off-hours task. >>  H >> Couple this with the CASA solution that allows you to virtualize yourH >> LUNs across controllers, even over IP WAN, and you've got some prettyG >> cool capabilities for high availability solution as well (far better $ >> than the EMA DRM solution, imho). > 0 >As long as we can get the throughput we need...  F As Rob has stated in another part of this discussion, that could be anC issue.  Not sure how your application stacks up in that requirement E area.  And, as he also pointed out, it's not in the "fully supported"  status yet either.   > G >> I think we have the technology available in different locations that D >> you can test-drive.  I think we also have some travelling "truck"? >> that's full of gear that goes around the country demoing our " >> technology.  Also, if you go toE >> DECUS/HPETS/whatever-its-called-this-year, you'll probably have an # >> opportunity for hands-on access.  > H >That'd be cool. For now, though, I'm still fighting off the "EMC is theG >cooling thing sinced sliced bread" crowd. (Too bad the damned thing is ? >unreliable with the shadow-sets that are the basis or Cerner's  >redundancy strategy.)  B I have worked with other health care offices who have been throughC similar discussions and analyses.  OpenVMS, Cerner app, and EVA  vs D EMC storage.  My own, personal, view into this is that the EMC costsE tend to be a lot higher, taking the next 3-5  years into account - it E can depend on how much your storage config needs to change over time. F We provide self-management of the storage at a somewhat cheaper price.   ------------------------------  % Date: Fri, 28 Feb 2003 09:53:39 -0500 & From: jlsue <jlsuexxxz@screaminet.com>$ Subject: Re: Volume set across RAID58 Message-ID: <lmtu5vo50a0oqpg1ht1bgope76s2sf214p@4ax.com>  E On Fri, 28 Feb 2003 06:38:48 -0500, John Santos <JOHN@egh.com> wrote:   . >On Thu, 27 Feb 2003, David J. Dachtera wrote: >[...] >  >>     > C >If you have a database disk you want to snapshot, I think what you A >do is take two partitions of the same size and shadow them (with < >traditional VMS volume shadowing).  When backup time comes,> >quiese the application, close all the files (or have it flushA >all pending I/O), split the shadow set by dismounting one member > >(or dismount the shadow set and remount just one member), and> >resume the application.  This should only take a few seconds.  F Although you can use HBVS this way, it's not guaranteed quite the sameA way if the LUNs are all on the same controller.  Controller-based C VRAID guarantee the redundancies aren't exposed to failure with the @ original data.  But HBVS can't do the same thing within the same controller.   > You can pretty easily - via DCL scripts  - create snapshots orE snapclones to automate this at the controller level.  It's much nicer B than the hszterm-like utilities of old when it comes to scripting.   ------------------------------   Date: 28 FEB 2003 16:21:13 GMT4 From: karcher@thuria.waisman.wisc.edu (Carl Karcher)$ Subject: Re: Volume set across RAID56 Message-ID: <28FEB03.16211369@thuria.waisman.wisc.edu>  8 In a previous article, John Santos <JOHN@egh.com> wrote:  G ->.. If you have a database disk you want to snapshot, I think what you B ->do is take two partitions of the same size and shadow them (with& ->traditional VMS volume shadowing).    C Keep in mind that the snapshot capability is an extra cost  license F option for the EVA. It's sold in various maximum storage capacities of' 1.6, 3.1, 6.2, 12.3 and 17.5 TB chunks.    --G -- Carl Karcher, Waisman Computing Services, Waisman Center, UW-Madison 7 --                karcher.nomorespam@waisman.wisc.edu      ------------------------------  + Date: Fri, 28 Feb 2003 15:08:10 +0100 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> G Subject: Re: What makes the VMS port to Itanium/IPF/whatever difficult? ; Message-ID: <01KSZ13ZT3O29FNHX8@sysdev.deutsche-boerse.com>   D > The truth is that when VMS was written, a lot of it was written inI > Macro-32. A *lot*.  Transposing that code into something else is a time * > consuming and very error prone process.  > I > That said, we have recently made the decision that not only will no new F > code be written in Macro-32 (C is the default for any new code), butF > that we will take every "reasonable" opportunity to rewrite Macro-32I > source into C.  But it will be case by case... since again, this can be J > tricky and time consuming.  There are some very specific areas of the OSI > where dumping the Macro-32 code would not only make maintenance easier, D > but will help performance.  By and large compilers optimize *much*B > better than hand coded assembler.  And even though Macro-32 is aH > compiler, because it is emulating an assembler it doesn't do as good a > job optimizing as it could.  > L > We are not rewriting Bliss code at this point.  But it has been raised (by, > me) as somthing to consider in the future.  F Perhaps you could comment on what was the reason why so many problems I crept into VMS MAIL (a program which I have spend several hours a day in  I for the past 12 years) when it was rewritten in C from BLISS.  Also, WHY  I was it rewritten?  VMS MAIL (for better or worse) is not an area where a  E lot of new functionality is being added (though the rewrite did take  I something away).  As a big fan of VMS MAIL and EDT, I often SPAWN out of  H EDT while in MAIL (there are several separate reasons I do this, and it > is also handy when one has only one session available).  What + specifically annoys me is the error message   9    command tables have invalid format - see documentation   E (that's right, no %DCL-E-TBLINV or whatever as a prefix).  Using the  F undocumented MAIL/OLD to run MAIL_OLD.EXE doesn't have this problem.  G (OK, it doesn't allow new stuff like SET SIGNATURE to work, but I have  I had EDT macros for stuff like that for a long time---and actually prefer  C them, since I can decide on a message-by-message basis with just a  G keystroke what stuff (signature, quote symbol, automatically generated  A quote du jour etc) I want added to a message).)  I mean, the old  I executable still ships and there is an extra (undocumented) qualifier to  F invoke it, which reeks of a pretty big screw up.  This has been years , ago and as far as I know is still not fixed.  H Most folks don't care what language VMS is written in.  But things like I this make one suspicious of rewrites, especially if it is a rewrite just  ? for the sake of a rewrite.  Also, it is a strength of VMS that  F applications can be written in any language with equal treatment from D the OS; the fact that C has a priviledged position in unix probably E comes from the fact that unix is written in C (and, I believe, C was  G written in order to write unix).  While this COULD still remain if all  H of VMS is written in C (why C, by the way---why not, say, Fortran 95?), ) there is a legitimate fear that it won't.   G Of course, slowly moving to a HLL is a good idea, so I'm not against a  I rewrite in general, but I've mentioned some reasons why folks tend to be   suspicious of rewrites.   & By the way, in case anyone is curious:  7 DEFINE KEY FUNCTION 29 AS "XLATESPAWN ?*'SPAWN '^ZREF."    ------------------------------  % Date: Fri, 28 Feb 2003 12:47:15 -0500 A From: "Fred Kleinsorge" <kleinsorge@star-dot-zko-dot-dec-dot-com> G Subject: Re: What makes the VMS port to Itanium/IPF/whatever difficult? . Message-ID: <3e5fa0a4$1_2@hpb10302.boi.hp.com>  F "Phillip Helbig" <HELBPHI@sysdev.deutsche-boerse.com> wrote in message5 news:01KSZ13ZT3O29FNHX8@sysdev.deutsche-boerse.com...  > G > Perhaps you could comment on what was the reason why so many problems  > crept into VMS MAIL   K Nope.  Not really.  I have no insight whatsoever into the internal workings  of mail.  J But you point out the problems with porting.  I don't know what the reasonH for the port was, or if it was justified.  But it's easy to believe thatD anything that isn't trivially small/simple will suffer *even* if the1 re-write is a transposition, and not a re-design.   I That is why we have always taken a very conservative approach as we moved H from VAX to Alpha to IPF.  But we have reached a point where some things9 need to be rewritten and designed for us to move forward.   K BTW - it is in many ways harder to verify the correctness of an interactive F program like MAIL than it is to say, verify that a complete rewrite ofF kernel spinlocks are correct.  I've spent a lot of time looking at X11I bugs - another one that despite the availability of verification and test L tools, has just too many combinations of events and paths to get all of them correct.   ------------------------------  % Date: Fri, 28 Feb 2003 12:16:37 +0000 - From: Roy Omond <Roy.Omond@BlueBubble.UK.Com> : Subject: Re: Wow - OpenVMS Job Advert in UK Computer Press/ Message-ID: <v5ukph49n5mvbf@corp.supernews.com>    Tim Llewellyn wrote:   > Simon Clubley wrote: > " > >In article , Roy Omond  writes: > > D > >>The position is in Cambridge, which would have been nice for me. >  > > H > >Actually, I'm more interested in knowing what the VMS market is like  > >in the North... :-) >  > G > There is a VMS Basic Programmer position in West Yorks that has been  4 > knocking around on jobserve for months and months.  B West Yorkshire is *not* the North.  I'd say the North begins northA of Perth and Dundee.  So if Simon really is looking for something B in the North (Inverness ? Ullapool ?  Dounray ? :-), I don't think he's going to get many offers.  	 Roy Omond  Blue Bubble Ltd.   ------------------------------    Date: 28 Feb 2003 07:10:08 -0600B From: clubley@remove_me.eisner.decus.org-Earth.UFP (Simon Clubley): Subject: Re: Wow - OpenVMS Job Advert in UK Computer Press3 Message-ID: <ct$XpW$Zbsug@eisner.encompasserve.org>   _ In article <v5ukph49n5mvbf@corp.supernews.com>, Roy Omond <Roy.Omond@BlueBubble.UK.Com> writes:  > Tim Llewellyn wrote: >  >> Simon Clubley wrote:  >> >> >I >> >Actually, I'm more interested in knowing what the VMS market is like   >> >in the North... :-)  >> >>H >> There is a VMS Basic Programmer position in West Yorks that has been 5 >> knocking around on jobserve for months and months.  > D > West Yorkshire is *not* the North.  I'd say the North begins northC > of Perth and Dundee.  So if Simon really is looking for something D > in the North (Inverness ? Ullapool ?  Dounray ? :-), I don't think  > he's going to get many offers. >   D I'm in North Yorkshire, in the general area of York, and I've always% considered myself to be in the North.   K I'm not actually looking to change jobs (so my boss's blood pressure can go I back down if he's reading this :-)), but I'm just trying to get a feeling > for VMS employment levels around the northern part of England.   Thanks to Tim for the pointer.   Simon.   --  B Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP       L VMS advocate: One who makes a Mac advocate look like a beginner at advocacy.   ------------------------------   End of INFO-VAX 2003.116 ************************