1 INFO-VAX	Wed, 13 Aug 2003	Volume 2003 : Issue 446       Contents:# Re: diff between SH LIC and PRODUCT + DLT FAST_SKIP problem with VMS 7.3-1 repost 1 Re: DQDRIVER  / OpenVMS 7.3-1 / ACER-Chipset bug?  Re: HP FUDBusting : Re: HP services in disarray ... IBM starting to look good?% Re: itrc - H.P. I.T. Resource Center.  Re: Netware is no VMS  Re: PDP-11 OS Release Dates  Re: PDP-11 OS Release Dates  Re: PDP-11 OS Release Dates  Re: PDP-11 OS Release Dates 3 RE: Sun tries to woo AlphaServer users away from HP ) Re: Upgrading VMS Cluster to DS10 or ES40 ) Re: Upgrading VMS Cluster to DS10 or ES40 ) Re: Upgrading VMS Cluster to DS10 or ES40 ) Re: Upgrading VMS Cluster to DS10 or ES40 ) Re: Upgrading VMS Cluster to DS10 or ES40   F ----------------------------------------------------------------------  # Date: Wed, 13 Aug 2003 07:18:10 GMT - From: "labadie" <tonari_no_tottoro@127.0.0.1> , Subject: Re: diff between SH LIC and PRODUCT2 Message-ID: <SCl_a.2125$0K3.1484@news.cpqcorp.net>  7 "amit sawhney" <thick_guy_9@yahoo.com> wrote in message : news:20030813010017.23296.qmail@web41104.mail.yahoo.com...	 > Hi all, 3 > Why don't I see exactly the same layered products 0 > using SHOW LICENSE and PRODUCT SHOW PRODUCT *? >   " You should have a look at the file" sys$update:vmsinstal_lmfgroups.com; which explains that a license can mean differents products.    Regards    Grard   Volem rien foutre al pais    ------------------------------  # Date: Wed, 13 Aug 2003 10:40:01 GMT + From: Jeff Cameron <JCam90502@jcameron.com> 4 Subject: DLT FAST_SKIP problem with VMS 7.3-1 repost2 Message-ID: <BB5F6391.A97E%JCam90502@jcameron.com>   REPOST: 
 ---------- To any who can help,  B In versions of VMS 7.2-2 and earlier the SCSI magnetic tape driverI SYS$MKDRIVER.EXE had code such that if an error or mount verification had D occurred on a tape device that supported fast skip, the driver wouldL permanently disable fast skip for that device, regardless of the setting setG by the $SET MAGTAPE/FAST_SKIP=option command. The only way to re-enable K fast_skip was to reboot the system. Now, Compaq/HP has provided a patch for # this problem which works just fine.   I But my question is this. In version 7.2-2 and earlier versions,one could L determine if fast skip for a tape device was disabled by doing the following :    $ANALYZE/SYSTEM  SDA>SHOW DEVICE <tape-device>  SDA>EXAMINE UCB+15C   L This would result in a hexadecimal long word. Looking at bit 12 of this wordL (identified by this mask %X'00001000'), if the bit is set, then fast_skip isG enabled, and if not set then fast_skip is permanently disabled. However J sometime between 7.2-2 and 7.3-1 this method of test seems to be no longerF valid, even though the native 7.3-1 SCSI magtape driver still disables* fast_skip on errors or mount verification.  L The patch kit dec-axpvms-vms731_fibre_scsi-v0300 does indeed fix the problem on VMS 7.3-1 (see : L http://ftp.support.compaq.com/patches/public/Readmes/vms/dec-axpvms-vms731_f ibre_scsi-v0300--4.README)  F Can anyone who has access to the source for the 7.3-1 SYS$MKDRIVER.EXEH provide a new test method to determine if fast skip has been disabled or not?   Thank you in advance.    Jeff Cameron   ------------------------------  % Date: Wed, 13 Aug 2003 09:37:37 +0200 $ From: "Peter Flunger" <p-i-b@gmx.at>: Subject: Re: DQDRIVER  / OpenVMS 7.3-1 / ACER-Chipset bug?9 Message-ID: <bhcps2$vvu48$1@ID-201992.news.uni-berlin.de>   E "Eberhard Heuser-Hofmann" <vaxinf@chclu.chemie.uni-konstanz.de> wrote  >  > Hi,  > K > There are several people who wrote me that they cannot use CDRECORD under K > OpenVMS 7.3-1. This only happens with ACER-Chipsets. The DQDRIVER-version F > is X-35A1. CDRECORD crashes during setting up the drive for burning. > < At the moment we are using a dqdriver we compiled and linked@ ourselves from dqdriver.c on the freeware CD because with X-35A1> we had all kinds of troubles when writing to CD or DVD burners# behind 'ACER 5229' IDE Controllers. ( The resulting sys$dqdriver image reports# 'image file identification: "X-26"'   and is working without problems. Peter    ------------------------------    Date: 13 Aug 2003 09:09:48 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)  Subject: Re: HP FUDBusting3 Message-ID: <x$XpVFg8Cax6@eisner.encompasserve.org>   Y In article <bh16ai$1pjj$4@jeeves.eng.abbnm.com>, peter@abbnm.com (Peter da Silva) writes:  > L > After all, the PDP-11 happens to be one that does. Split I&D and pure textF > code in V7 and 2BSD had non-executable stacks by default... but 4BSDI > didn't, and running 4BSD is what most RISCs seem to have been optimised  > for.  D    RSX-11M also split code into I and D space, and under that I haveD    executed code on the stack.  Splitting up code into I and D spaceD    on the PDP-11, or pure and impure on the PDP-10, or EXE and NOEXEG    PSECTs on VAXen and Alpha, does NOT prevent one from jumping to the  D    stack and executing code there.  These processors don't care whatG    address is in the PC as long as it's mapped to a readable address as G    they don't have, or don't fully implement, a no-execute bit in their %    hardware level of page protection.    ------------------------------    Date: 13 Aug 2003 09:46:58 -0700( From: bob@instantwhip.com (Bob Ceculski)C Subject: Re: HP services in disarray ... IBM starting to look good? = Message-ID: <d7791aa1.0308130846.328c6459@posting.google.com>   m bob@instantwhip.com (Bob Ceculski) wrote in message news:<d7791aa1.0308120919.7426d30e@posting.google.com>... 6 > this is on the inquirer ... if this is true, then HP > is in trouble big time!  > + > http://www.theinquirer.net/?article=10950   ! this doesn't sound any better ...   ) http://www.theinquirer.net/?article=10996    ------------------------------  % Date: Wed, 13 Aug 2003 00:20:22 -0400 - From: Jay Newman <fredthecat@i-love-cats.com> . Subject: Re: itrc - H.P. I.T. Resource Center.8 Message-ID: <pf6jjv4of1eg1b2bjau7a1n8k1ibqpps9b@4ax.com>  M I have to add my voice to the disappointment I have heard on this thread over N the last day. It does appear that HP is reducing functionality & accessibilityM without providing much in the line of improvements to make it worth while (or  make sense.)  P However, here is a "plus" I have seen. One of the other VMS admins at my shop isN much better versed at HP UX, and quite familiar with the ITRC system. He is inM the middle of upgrading the O/S on an Alpha to VMS 7.3-1, and found it simple B enough to navigate & come up with the list of patches to download.7 Not bad for someone relatively unfamiliar with OpenVMS.   J If we're lucky, this kind of approach (using the ITRC for OpenVMS as it isL already used for Unix admins etc.) may help OpenVMS lose some of its aura as' being obscure and difficult to manage.    . - A non-marketing approach to marketing VMS ??  B ... or just another attempt to put a round peg into a square hole? Time will tell.    ------------------------------    Date: 13 Aug 2003 11:47:08 -0500+ From: young_r@encompasserve.org (Rob Young)  Subject: Re: Netware is no VMS3 Message-ID: <n2okXPDvRWD0@eisner.encompasserve.org>   a In article <7pnfjv0kt2vc82iba3lh0c6b1qj93kvuc7@4ax.com>, jlsue <jefflsxxxz@sbcglobal.net> writes: L > On 6 Aug 2003 23:31:58 -0500, young_r@encompasserve.org (Rob Young) wrote: > f >>In article <bgrube$po1nb$1@ID-120847.news.uni-berlin.de>, "John Travell" <john@jomatech.com> writes:7 >>> "jlsue" <jefflsxxxz@sbcglobal.net> wrote in message 6 >>> news:ueb2jvs5e4upk428af4h2orv3bu1j3vvfg@4ax.com...4 >>>> I know this take us a little off-topic, but.... >>>>L >>>> What new features do you believe are needed above what's already in theI >>>> roadmap?  For me, at this point, I don't see any real missing killer M >>>> feature that's both feasible (i.e., not something like process migration * >>>> between cluster nodes), and valuable. >>>> >>>  >>A >>	Maybe in a few days a few good ideas surface.  I guess you are A >>	saying all the good ideas for OS improvement have already been  >>	invented?  Of course not. > I > Right.  Of course not.  That's why I stated it as a question, wondering I > what people would like.  But I did add that I don't foresee any "killer ? > feature" yet.  I remain hopeful that one will arise, however.  >  >> [snip TRAM and MRAM]  > I > But these aren't killer OS features.  These are memory-related, and all ? > that's needed is OS support.  In my mind these are different.  >  >>H >>	The missing killer VMS feature is true WRITEBEHIND (with confidence) @ >>	at a filesystem level.  With MRAM, the writes are not lost on; >>	power outtage.  So as Fred mentioned what the impetus or # >>	catalyst for Galaxy's birth was:  > I > Well, that's always a possibility... once the technology for truly safe H > writing is in place.  I'm not holding my breath, however, because very2 > often these promises don't see the light of day. >   : 	You ripped my premise that ensures MRAM not only shows up< 	but will show up.  Part of such a strong statement is based= 	on the technology - i.e. 15% faster than DRAM (read a link).   9 	Secondly, MRAM maintains state on power down.  You can't  	argue against that at all.   = 	So Galaxy developers "saw" large (or big systems as Fred K.  F 	so eloquently states) coming.  How to make use of large systems with - 	NUMA characteristics?  Galaxy - of course.     6 	What will the response be to MRAM?  Same old same-oh?   				Rob       ------------------------------  ! Date: Wed, 13 Aug 03 09:48:53 GMT  From: jmfbahciv@aol.com $ Subject: Re: PDP-11 OS Release Dates+ Message-ID: <bhd5ac$l8g$1@bob.news.rcn.net>   . In article <mdd4r0mzr6l.fsf@panix5.panix.com>,7    Rich Alderson <news@alderson.users.panix.com> wrote:  >jmfbahciv@aol.com writes: > 6 >> In article <D2JkWN3WpxD9@eisner.encompasserve.org>,B >>    koehler@eisner.nospam.encompasserve.org (Bob Koehler) wrote: > I >>> In article <bh88d7$qb3$1@bob.news.rcn.net>, jmfbahciv@aol.com writes:  > E >>>> And were all files limited to one length?  What do you do if you D >>>> have a file that is larger than available pre-allocated memory? > G >>> You used files as files.  You read a record or two, processed that, B >>> and wrote a record or two to the output file.  You didn't needF >>> a 256K of memory just because you had a file that took up the half >>> the floppy.  > D >> That's what sane people would have done :-).  I get the distinct * >> impression that MS-DOS didn't know how. > 8 >Ahh.  Your distinct impression is wrong in that case.   >There are other reasons6 >that MS-DOS loses readily, but that's not one of 'em.  @ Really?  Here's one reason I thought that way.  When I'm editingA a file and exceed some flavor of size limit, the computer freezes @ up.  It used to mess up serious stuff but somehow (by den motherB accidental magic) I managed to have the keyboard code get affected@ before disk code.  That way I can't type in anymore and mess the@ FAT.  A short-byte way around this asspect is to exit and reloadA the file again.  Apparently it gets read in from top-down end-to- ' beginning--another fine mess potential.   @ There were other "examples" that fit the hypothesis, but I can't recall them now.  D I haven't managed to figure out a workaround if Windows eats itself.# That critter always blasts the FAT.    /BAH    ' Subtract a hundred and four for e-mail.    ------------------------------  ! Date: Wed, 13 Aug 03 09:55:42 GMT  From: jmfbahciv@aol.com $ Subject: Re: PDP-11 OS Release Dates+ Message-ID: <bhd5n5$l8g$2@bob.news.rcn.net>   ) In article <tqmekzqkfpl.fsf@drizzle.com>, ,    Patrick Scheible <kkt@drizzle.com> wrote: >jmfbahciv@aol.com writes: > , >> In article <tqmy8y088ka.fsf@drizzle.com>,/ >>    Patrick Scheible <kkt@drizzle.com> wrote:  >> >jmfbahciv@aol.com writes:  >> >5 >> >> In article <bh5q7m$ddm$1@tabloid.uwaterloo.ca>, 8 >> >>    dfevans@bcr10.uwaterloo.ca (David Evans) wrote:H >> >> >In article <bh58k6$mlg$2@bob.news.rcn.net>,  <jmfbahciv@aol.com>  wrote:7 >> >> >>In article <bgtj2h$i34$1@tabloid.uwaterloo.ca>, : >> >> >>   dfevans@bcr10.uwaterloo.ca (David Evans) wrote:J >> >> >>>In article <bgtate$60n$4@bob.news.rcn.net>,  <jmfbahciv@aol.com> 	 >> wrote: < >> >> >>>>In article <3f3f0cd2.12775862@news.supernews.com>,- >> >> >>>>   ian@hammo.com (paramucho) wrote: L >> >> >>>>>One of the big problems with cramped spaces was that users would  >> useK >> >> >>>>>up every available byte of user space available, thus increasing > >> >> >>>>>system size would render their programs inoperable.
 >> >> >>>>H >> >> >>>>This makes no sense.  That would be problem only if their codeB >> >> >>>>(or data placement) was hardwired to absolute addresses.
 >> >> >>>>	 >> >> >>> H >> >> >>>  In my experiences in micro land, absolute addresses were the  rule. K >> >> >>>Not many folks bothered with relocating loaders when you only had   one F >> >> >>>64K address space and something like the 6502 makes it pretty= >> >> >>>irritating to write fully position-independant code.  >> >> >>E >> >> >>Wait a minute...this doesn't make any sense.  Code had to be   locationJ >> >> >>independent if it can be loaded into core starting at any address. >> >> > L >> >> >  That was my point: frequently one could not load code into core at  >> anyL >> >> >address and expect it to work.  The assembler would be provided with  >> theL >> >> >expected start address and resolve labels to absolute addresses.  As  >> thereG >> >> >was no proper link loader (loading the code simply consisted of   >> stuffing J >> >> >the bytes in memory beginning at the proper address) no relocation  >> could >> >> >be done.  >> >> I >> >> Oh, gagmeverymuch.  No wonder there's code bloat and memory leakage G >> >> out there.  This means that a program had to know how much memory D >> >> it ever wanted to use before assembly.  There was no mechanismE >> >> to ask the monitor for more core (address space) if and when it E >> >> needed it.  Most programmers would issue an over-inflated BLOCK D >> >> statement just to play it "safe" and never bother to do checks, >> >> for overflowing its own address space. >> > >> >E >> >On the weenie machines, there only one program running at any one 
 >> >time.  >>  " >> Some days, yes; some days, not. > F >Pretty much all days.  It was a Great Leap Forward when just printing' >didn't completely occupy the computer.   ? Nah, think about front ends handling comm and output.  -8s were < very, very good at handling a certain task without requiring= human babysitting.  And those marvels could work in all kinds " of environments without a <CTRL>G.    A >> No, it can't.  It does have know where it's own code stops and = >> starts.  Every program has some memory management, even if  >> it's implicit.  > < >Right, but the program only had to manage memory to its own >satisfaction.    ? Sure. :-) I get this strange feeling that programmers out there  don't do this.  5 > ..It didn't have to worry about being nice to other $ >programs because there weren't any.  2 There is always sensitive spots in a computer. ;-) > 2 >> > ..If it wanted more memory than there was, inF >> >most cases it had to figure out how to store some of it on disk or< >> >floppy all by itself without help from pager or the OS.  >>  9 >> Sure.  Most programs already have this; I guess people + >> don't recognize it as memory management.  > : >It's different from managing memory in a multitasking OS.  > I'm not even trying to imply that being an overseer isn't more= complicated.  I've been talking about people who aren't aware 9 that such things have to exist, even in stand-alone mode. > I seem to sense an attitude that the "compiler" will take care; of it.  That's exactly the wrong place.  The "compiler" has ( nothing to do with it at execution time.   <snip>   /BAH    ' Subtract a hundred and four for e-mail.    ------------------------------  ! Date: Wed, 13 Aug 03 09:58:23 GMT  From: jmfbahciv@aol.com $ Subject: Re: PDP-11 OS Release Dates+ Message-ID: <bhd5s5$l8g$3@bob.news.rcn.net>   . In article <mdd7k5izrfb.fsf@panix5.panix.com>,7    Rich Alderson <news@alderson.users.panix.com> wrote: , >bob smith <sfmc68@bellatlantic.net> writes: > H >> 5u60 is the small kind cassette tape, tu58 was DECTape/LincTape size > >> spools, and the TU10 stuff was the industrial size magtape. > L >TU-58 was microcassette--I know, I loaded enough of them into the HSC-50's  on: >our CI cluster at LOTS.  DECtape/LincTape is TU-55/TU-56.  = Yes, I would have shot the product management who deemed that > the crap cassette was even remotely like our beloved DECtapes.4 By that time, we figured everybody north was nuts.     /BAH    ' Subtract a hundred and four for e-mail.    ------------------------------    Date: 13 Aug 2003 09:59:24 -0700( From: Patrick Scheible <kkt@drizzle.com>$ Subject: Re: PDP-11 OS Release Dates) Message-ID: <tqmvft1334j.fsf@drizzle.com>    jmfbahciv@aol.com writes:   A > Nah, think about front ends handling comm and output.  -8s were > > very, very good at handling a certain task without requiring? > human babysitting.  And those marvels could work in all kinds $ > of environments without a <CTRL>G.  C I never got to use an -8.  Was that a hardware front end or part of  the OS?   A > Sure. :-) I get this strange feeling that programmers out there  > don't do this.  ? Some of their standards for satisfaction are pretty low.  If it F doesn't crash so often that the customer takes it back within 30 days, it's good...  @ > I'm not even trying to imply that being an overseer isn't more? > complicated.  I've been talking about people who aren't aware ; > that such things have to exist, even in stand-alone mode. @ > I seem to sense an attitude that the "compiler" will take care= > of it.  That's exactly the wrong place.  The "compiler" has * > nothing to do with it at execution time.  
 -- Patrick   ------------------------------  % Date: Wed, 13 Aug 2003 08:02:06 -0700 # From: "Tom Linden" <tom@kednos.com> < Subject: RE: Sun tries to woo AlphaServer users away from HP9 Message-ID: <CIEJLCMNHNNDLLOOGNJIIEECHMAA.tom@kednos.com>   E The HP/compaq's treatment of PL/I certainly isn't founded on any kind B of defensible business practice, but on the other hand, since your; company has not paid for PL/I support in more than 5 years, > (during which time we have issued two releases of the product,< made all documentation and kits available for free download)" you have little to complain about.     >-----Original Message----- 5 >From: Jeff Goodwin [mailto:jgoodwin@maine.rrr-r.com] ' >Sent: Wednesday, July 30, 2003 5:32 AM  >To: Info-VAX@Mvb.Saic.Com= >Subject: Re: Sun tries to woo AlphaServer users away from HP  >  >  > E >I got a call from Sun a couple of weeks ago concerning HP's mythical L >announcement that OpenVMS had been EOL'd.  By now, I suspect most remaining9 >OpenVMS customers are on to this re-occurring deception.  > A >Unfortunately, it will be Digital/Compaq/HP's pathetic treatment  >of the PL/1I >compiler that will eventually drive our company to another vendor.  (No, L >we're not converting it to C++ so we can run on IPF.  If we have to perform> >a large-scale conversion, it'll be away from the company that >forced us into K >the conversion).  Based on what other divisions are doing in this company, / >there's a good chance that vendor will be Sun.  >  >-Jeff > ? >"Keith Parris" <keithparris_NOSPAM@yahoo.com> wrote in message 8 >news:cf15391e.0307291812.3327f013@posting.google.com...G >> In a campaign called 'HP Away', Sun is trying to take customers from G >> HP, but based on the customers quoted in these articles, they aren't  >> having much luck. >> >[snip]  >  >  >---' >Incoming mail is certified Virus Free. ; >Checked by AVG anti-virus system (http://www.grisoft.com). A >Version: 6.0.502 / Virus Database: 300 - Release Date: 7/18/2003  >  --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).? Version: 6.0.506 / Virus Database: 303 - Release Date: 8/1/2003    ------------------------------  # Date: Wed, 13 Aug 2003 04:06:15 GMT % From: "Mike Naime" <mnaime@kc.rr.com> 2 Subject: Re: Upgrading VMS Cluster to DS10 or ES40; Message-ID: <XOi_a.88194$o27.1937989@twister.rdc-kc.rr.com>   ; >Grealy, Patrick <PGrealy@sph.uth.tmc.edu> wrote in message G news:40B30BA5E139F648B2EA60E6AE3A7B2601CCA278@sphnt3.sph.uth.tmc.edu... E >Our current cluster is DEC7610 + DEC2100 + HSJ40/StorageWorks. We're H considering the cheapest way to migrate to 21st century and maintain theH >redundancy of the cluster. Any suggestions considering DS10 vs. ES40? - Thanks.   8 1.) How much memory do you want to install into the box?7 2.) How many internal PCI cards do you need to install? J 3.) Are you installing internal storage?  Or is it all the external HSJ's.   DS10: 0 One CPU.  Thats it!  No room for expansion here.6 Only 4 PCSI slots.  This may not be a problem for you.I 2 built in internal 10-100 network ports.  (WE0, WE1)  If you do not need  more than this, this is a plus. ! You can only put 2GB in the DS10. & Space for a couple of internal drives.. ONLY ONE POWER CORD.  SINGLE POINT OF FAILURE!    C The ES40 has more room for later expansion than you do in the DS10.  10 PCSI slots on 2 PCSI buses.D Up to 4 CPU's.  You do not need to buy all of them at the same time.! Space for 2 internal drive cages. J You can install up to 32GB of memory into the ES40 if you have all 3 power supplies installed. 2 3 power supplies allows for redundant power feeds.  H An ES40 with 4 CPU's and 8GB of memory CAN run on only one power supply.I (No internal disk drives)  I have had at least 3 power supply failures in K the last 2 years in my ES4x servers.  We were able to hot swap/replace with  no downtime!  F I would recommend the ES40 over the DS10.  Even if you bought the sameI amount of memory and CPU.  With the ES40, you could add CPU/RAM next year ( without having to buy a totally new box.   Check Ebay! @ $ 9K for -- CPU: 4x667mhz    Memory: 1GB RAM    Disk: 4x9.1GB HD% OR a 1CPU 833Mhz w 1GB ram for $11.5K 5 you might find that used ES40 cheaper than you think.   J Note: if your processes are I/O hogs, consider fibre channel storage, thatB will really boost your I/O throughput more than a bigger box will!   ------------------------------  # Date: Wed, 13 Aug 2003 13:43:33 GMT % From: "Mike Naime" <mnaime@kc.rr.com> 2 Subject: Re: Upgrading VMS Cluster to DS10 or ES40; Message-ID: <9gr_a.89998$o27.1957950@twister.rdc-kc.rr.com>   / Island <dbturner@islandco.com> wrote in message ) news:vjkagoilnql29e@news.supernews.com... " > What about DS20 or DS20e systems > L > DS20 - 2 x 500Mhz CPU (or higher -works but not supported) Dual P/Supply 7- > Disk Slots- Up to 4GB  Cheaper than an ES40 L > DS20e 2 x 500-833Mhz CPU, 6 SLot Disk Cage, 3 Power Supplies (minimum 2 to* > run system)Up to 4GB Cheaper than a ES40 >   I Yes, The DS20 does have multiple power supplies internally,  But the DS20 ) still has only a single power cord on it. B I was trying to answer his question on comparing the 10 to the 40.   Oh yea...   Another point.  L DS10 -- you need to take it out of the rack and set it on a table to work onJ it.  The rack mount kit does not enable you to work on it in the rack.  NoI cable arm assembly.  Just something to slide it into the rack and hold it J there.  It is very easy to work on in that everything latches/slides apart without any tools.  L ES40 -- While rack mounted, you can replace just about everything except for5 the PCI backplane without detaching it from the rack. @ You must have the correct hex key to open/close the panel doors.   >  >  > DT >  > -- > David B Turner! > Island Computers US Corporation  > 2700 Gregory St., Suite 180  > Savannah GA 31404  > Tel: 912 447 6622  > Fax: 912 201 0402  > Email: dbturner@hpaq.net > http://www.hpaq.net  >  >  > 2 > "Mike Naime" <mnaime@kc.rr.com> wrote in message7 > news:XOi_a.88194$o27.1937989@twister.rdc-kc.rr.com...  > > ? > > >Grealy, Patrick <PGrealy@sph.uth.tmc.edu> wrote in message K > > news:40B30BA5E139F648B2EA60E6AE3A7B2601CCA278@sphnt3.sph.uth.tmc.edu... I > > >Our current cluster is DEC7610 + DEC2100 + HSJ40/StorageWorks. We're L > > considering the cheapest way to migrate to 21st century and maintain theL > > >redundancy of the cluster. Any suggestions considering DS10 vs. ES40? - > > Thanks.  > > < > > 1.) How much memory do you want to install into the box?; > > 2.) How many internal PCI cards do you need to install? G > > 3.) Are you installing internal storage?  Or is it all the external  HSJ's. > > 	 > > DS10: 4 > > One CPU.  Thats it!  No room for expansion here.: > > Only 4 PCSI slots.  This may not be a problem for you.H > > 2 built in internal 10-100 network ports.  (WE0, WE1)  If you do not need# > > more than this, this is a plus. % > > You can only put 2GB in the DS10. * > > Space for a couple of internal drives.2 > > ONLY ONE POWER CORD.  SINGLE POINT OF FAILURE! > >  > > G > > The ES40 has more room for later expansion than you do in the DS10. " > > 10 PCSI slots on 2 PCSI buses.H > > Up to 4 CPU's.  You do not need to buy all of them at the same time.% > > Space for 2 internal drive cages. H > > You can install up to 32GB of memory into the ES40 if you have all 3 power  > > supplies installed. 6 > > 3 power supplies allows for redundant power feeds. > > L > > An ES40 with 4 CPU's and 8GB of memory CAN run on only one power supply.J > > (No internal disk drives)  I have had at least 3 power supply failures inJ > > the last 2 years in my ES4x servers.  We were able to hot swap/replace > with > > no downtime! > > J > > I would recommend the ES40 over the DS10.  Even if you bought the sameH > > amount of memory and CPU.  With the ES40, you could add CPU/RAM next year, > > without having to buy a totally new box. > >  > > Check Ebay! D > > $ 9K for -- CPU: 4x667mhz    Memory: 1GB RAM    Disk: 4x9.1GB HD) > > OR a 1CPU 833Mhz w 1GB ram for $11.5K 9 > > you might find that used ES40 cheaper than you think.  > > I > > Note: if your processes are I/O hogs, consider fibre channel storage,  thatF > > will really boost your I/O throughput more than a bigger box will! > >  > >  > >  > >  > >  > >  >  >    ------------------------------  % Date: Wed, 13 Aug 2003 08:05:59 -0400 & From: "Island" <dbturner@islandco.com>2 Subject: Re: Upgrading VMS Cluster to DS10 or ES40/ Message-ID: <vjkagoilnql29e@news.supernews.com>     What about DS20 or DS20e systems  J DS20 - 2 x 500Mhz CPU (or higher -works but not supported) Dual P/Supply 7+ Disk Slots- Up to 4GB  Cheaper than an ES40 J DS20e 2 x 500-833Mhz CPU, 6 SLot Disk Cage, 3 Power Supplies (minimum 2 to( run system)Up to 4GB Cheaper than a ES40       DT   --   David B Turner Island Computers US Corporation  2700 Gregory St., Suite 180  Savannah GA 31404  Tel: 912 447 6622  Fax: 912 201 0402  Email: dbturner@hpaq.net http://www.hpaq.net       0 "Mike Naime" <mnaime@kc.rr.com> wrote in message5 news:XOi_a.88194$o27.1937989@twister.rdc-kc.rr.com...  > = > >Grealy, Patrick <PGrealy@sph.uth.tmc.edu> wrote in message I > news:40B30BA5E139F648B2EA60E6AE3A7B2601CCA278@sphnt3.sph.uth.tmc.edu... G > >Our current cluster is DEC7610 + DEC2100 + HSJ40/StorageWorks. We're J > considering the cheapest way to migrate to 21st century and maintain theJ > >redundancy of the cluster. Any suggestions considering DS10 vs. ES40? -	 > Thanks.  > : > 1.) How much memory do you want to install into the box?9 > 2.) How many internal PCI cards do you need to install? L > 3.) Are you installing internal storage?  Or is it all the external HSJ's. >  > DS10: 2 > One CPU.  Thats it!  No room for expansion here.8 > Only 4 PCSI slots.  This may not be a problem for you.K > 2 built in internal 10-100 network ports.  (WE0, WE1)  If you do not need ! > more than this, this is a plus. # > You can only put 2GB in the DS10. ( > Space for a couple of internal drives.0 > ONLY ONE POWER CORD.  SINGLE POINT OF FAILURE! >  > E > The ES40 has more room for later expansion than you do in the DS10.   > 10 PCSI slots on 2 PCSI buses.F > Up to 4 CPU's.  You do not need to buy all of them at the same time.# > Space for 2 internal drive cages. L > You can install up to 32GB of memory into the ES40 if you have all 3 power > supplies installed. 4 > 3 power supplies allows for redundant power feeds. > J > An ES40 with 4 CPU's and 8GB of memory CAN run on only one power supply.K > (No internal disk drives)  I have had at least 3 power supply failures in H > the last 2 years in my ES4x servers.  We were able to hot swap/replace with > no downtime! > H > I would recommend the ES40 over the DS10.  Even if you bought the sameK > amount of memory and CPU.  With the ES40, you could add CPU/RAM next year * > without having to buy a totally new box. > 
 > Check Ebay! B > $ 9K for -- CPU: 4x667mhz    Memory: 1GB RAM    Disk: 4x9.1GB HD' > OR a 1CPU 833Mhz w 1GB ram for $11.5K 7 > you might find that used ES40 cheaper than you think.  > L > Note: if your processes are I/O hogs, consider fibre channel storage, thatD > will really boost your I/O throughput more than a bigger box will! >  >  >  >  >  >    ------------------------------  % Date: Wed, 13 Aug 2003 09:11:43 -0500 ( From: brandon@dalsemi.com (John Brandon)2 Subject: Re: Upgrading VMS Cluster to DS10 or ES401 Message-ID: <03081309114304@dscis6-0.dalsemi.com>   H > > Our current cluster is DEC7610 + DEC2100 + HSJ40/StorageWorks. We'reL > > considering the cheapest way to migrate to 21st century and maintain theK > > redundancy of the cluster. Any suggestions considering DS10 vs. ES40? -  > > Thanks.  > M > ES40.  You can put a CIPCA-BA (PCI-CI port) in the ES40, but you can't put  3 > one in a DS10 (or, at least, it's not supported).  > M > Oh, by the way, when your 21st century friends start asking "Why don't you  I > upgrade to SAN", you can say that you've been using one ever since you  > > started using your HSJ40, and that they work, and work well. > L > At our shop, we've been using SAN technology since installing HSC50 (yes, " > that's hsC50) over 15 years ago.	 >     Stu      ES40 or better.   N We have a cluster of DS20's and ES40 and are at the crossroads of upgrades forL performance reasons - pain in the ars.  The DS20 very limited - by reason of" default so the DS10 would also be.  P You can, as Stu suggests, stay on your current SAN.  And that will be a low costH SAN.  However, the CI will pale in performance comparison to the HSG80'sI coupled with either ESA (old and cheaper) or the EMA (new).  Kind of like L putting an IRL on a 1/8 oval dirt track.  It will be the fastest race car on. the track but it will not be able to prove it!  N We migrated from a cluster of VAX 63xx & 76xx and A/S 2100 to a GS160 and usedN the CI as an interface between the old and new.  We ran HSJ40 & HSJ50's in ourI CI.  Though we moved the data from CI to Fiber we did see the comparitive O lackluster performance of the CI against the Fiber.  Big difference.  Maybe you J should consider the Fiber Phase II.  But I would push for its replacement.     J*o*h*n B*r*a*n*d*o*n  VMS Systems Administrator * firstname.lastname.spam.me.not@dalsemi.com   ------------------------------  % Date: Wed, 13 Aug 2003 12:49:41 -0400 & From: "Island" <dbturner@islandco.com>2 Subject: Re: Upgrading VMS Cluster to DS10 or ES40/ Message-ID: <vjkr4pb9qdv13c@news.supernews.com>   G Ebay is a great source for equipment until it stops working after a few  days. * $2000 for a Coffee table is extremely high   ;0)    David @ island    > "Malcolm Dunnett" <nothome@spammers.are.scum> wrote in message' news:KiZ7kVYRM2ol@malvm7.mala.bc.ca.... J > In article <9gr_a.89998$o27.1957950@twister.rdc-kc.rr.com>, "Mike Naime" <mnaime@kc.rr.com> writes: > > 3 > > Island <dbturner@islandco.com> wrote in message - > > news:vjkagoilnql29e@news.supernews.com... % > >> What about DS20 or DS20e systems  > >>D > >> DS20 - 2 x 500Mhz CPU (or higher -works but not supported) Dual
 P/Supply 70 > >> Disk Slots- Up to 4GB  Cheaper than an ES40L > >> DS20e 2 x 500-833Mhz CPU, 6 SLot Disk Cage, 3 Power Supplies (minimum 2 to- > >> run system)Up to 4GB Cheaper than a ES40  > >> > > H > > Yes, The DS20 does have multiple power supplies internally,  But the DS20- > > still has only a single power cord on it.  > J >    The DS20 has two power cords ( one for each P/S ). The DS20E only has> > one cord. The ES40 has one power cord for each power supply. > H >    DS20s ( 2 x 500Mhz CPU, 1GB memory ) have been seen to sell on Ebay > for well under $2000.I >nI >    Another consideration for many is license costs. Software for a DS10pJ > is much cheaper than software for an ES40. I believe the original poster was F > at an educational institution, so if he's on the EDU license program( > and/or CSLG that may not be a concern. > H >    If you can afford an ES40 ( and license costs aren't an issue ) I'd8 > go that way - it offers a lot more room for expansion. >s > >> >L > >> > Note: if your processes are I/O hogs, consider fibre channel storage, > > thatI > >> > will really boost your I/O throughput more than a bigger box will!e > >> > >rK >    I'd second that. The FC widgets can be found quite cheap on Ebay these- daysD > (KGPSAs for $20-500 each depending on the version, FC switches forA > $500-2000, RA8000s for a few thousand $ ). I can't speak for CI: performancewI > as I've never had one, but some simple throughput tests I did showed an K > RA8000 (FC) to have 4 to 5 times the data transfer speed of an RA7000 (FW1 SCSI).J > The downside is that SCS isn't currently carried over FC ( though rumourK > has it that may change some day ), so you'd need to keep your CI adapters E > or get a different interconnect, such as Ethernet, for your clusterm
 managementI > communications. OTOH if you're thinking of replacing your whole cluster  with* > a single ES40 that may not be a concern. >-   ------------------------------   End of INFO-VAX 2003.446 ************************