1 INFO-VAX	Sun, 13 Nov 2005	Volume 2005 : Issue 633       Contents:! Re: A question about autocomplete ! Re: A question about autocomplete  Re: DECwindows Startup Error? RE: failSAFE IP - Looks good! Who's using it? (and why/why-not) ? Re: failSAFE IP - Looks good! Who's using it? (and why/why-not) ? Re: failSAFE IP - Looks good! Who's using it? (and why/why-not) G Re: Lack of response on c.o.v. (was:Re: File spec wildcard match test?) G Re: Lack of response on c.o.v. (was:Re: File spec wildcard match test?)  Re: Microvax image to simh? - re: Request for feedback - BACKUP enhancement - Re: Request for feedback - BACKUP enhancement - Re: Request for feedback - BACKUP enhancement % Speaking of "fulcrumizing the foible" % Re: Strange Network performance issue 1 Re: Taking backup of disks of production system.. 1 RE: Taking backup of disks of production system.. 1 Re: Taking backup of disks of production system.. 1 Re: Taking backup of disks of production system.. 1 Re: Taking backup of disks of production system.. 1 Re: Taking backup of disks of production system.. = Re: Video card Digital PCI ZLXp-L1 compatibility with OpenVMS = Re: Video card Digital PCI ZLXp-L1 compatibility with OpenVMS G Re: Will Rich Marcello either come clean or do some proper marketingand G Re: Will Rich Marcello either come clean or do some proper marketingand G RE: Will Rich Marcello either come clean or do some proper marketingand G Re: Will Rich Marcello either come clean or do some proper marketingand G Re: Will Rich Marcello either come clean or do some proper marketingand G Re: Will Rich Marcello either come clean or do some proper marketingand O Re: Will Rich Marcello either come clean or do some proper marketingand PR PRPR P Re: Will Rich Marcello either come clean or do some proper marketingandPR market  F ----------------------------------------------------------------------  # Date: Sat, 12 Nov 2005 18:56:11 GMT ( From: Alan Greig <greigaln@netscape.net>* Subject: Re: A question about autocomplete= Message-ID: <fVqdf.22806$Xz1.16852@fe2.news.blueyonder.co.uk>     Peter 'EPLAN' LANGSTOEGER wrote:\ > In article <dl4n4r$cgj$1$8300dec7@news.demon.co.uk>, issinoho <issinoho@gmail.com> writes: > J >>I've always felt that one of the biggest failings with DCL was the lack J >>of a Unix-style autocompletion of filepaths when TAB is hit. This would B >>be a fantastic addition, however if I remember rightly an older I >>discussion around this offered some fundamental reasons why this would   >>never work under DCL.  >  > J > There is/was a freeware which did/does it. I saw it in the V5 timeframe. > Maybe it is/was VAX only...   D DCLCOMPLETE. It is VAX only. Works through current VAX/VMS 7.3. The D irony is that Unix copied this from DEC's TOPS-20. In the Goals and C Non-Goals for VMS this feature was a "non-goal". The TOPS-20 CMND%  I operating system call that implemented this functionality was recoded in   C for C-Kermit.    --  
 Alan Greig   ------------------------------  % Date: Sat, 12 Nov 2005 22:11:22 -0500 4 From: "Peter Weaver" <newsgroup@weaverconsulting.ca>* Subject: Re: A question about autocomplete9 Message-ID: <z9ydf.8961$Tk1.170433@news20.bellglobal.com>   0 "issinoho" <issinoho@gmail.com> wrote in message. news:dl4n4r$cgj$1$8300dec7@news.demon.co.uk...I > I've always felt that one of the biggest failings with DCL was the lack I > of a Unix-style autocompletion of filepaths when TAB is hit. This would A > be a fantastic addition, however if I remember rightly an older H > discussion around this offered some fundamental reasons why this would > never work under DCL.  >...  G Does nobody here keep an eye things coming out of the OpenOffice group?   # http://www.oooovms.dyndns.org/auto/    ------------------------------  % Date: Sat, 12 Nov 2005 17:54:18 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> % Subject: Re: DECwindows Startup Error , Message-ID: <43767297.ADFF68CD@teksavvy.com>   Ross Templeton wrote: 3 > DECW$TRANSPORT_TCPIP image base address: 005C4000 4 > %DECW-W-ATT_FAIL, failed to attach transport TCPIP+ > -SYSTEM-F-NOLOGNAM, no logical name match     > DECwindows gets started under the hood after systartup_VMS.com  ; In my systarttup_VMS, before the end, I have the following:   	 $!******* 	 $WAITTCP: 	 $!******* # $     if f$getdvi("_bg0:","exists") 
 $     then$ $         if f$getdvi("_bg0:","mnt") $         thenG $             say "TCPIP sufficiently started, continuing (''loopcnt')"  $             goto ENDWAIT $         endif  $     endif  $! $     wait 00:00:10  $     loopcnt = loopcnt + 1  $     say "Waiting..."* $     if loopcnt .lt. 18 then goto WAITTCP $!	 $ENDWAIT:     E Basically, it waits for the BG device to become available, indicating  TCPIP is sufficiently started.  D In my case, I submit a whole bunch of stuff, like tcpip startup to aC batch queue that has a job limit of 1.  The above code just "hangs" 7 systartup_vms.com until TCPIP is available, after which D systartup_vms.com ends, and the system proceeds with the rest of the startup including decwindows.    ------------------------------  % Date: Sat, 12 Nov 2005 23:25:41 -0500 ' From: "Main, Kerry" <Kerry.Main@hp.com> H Subject: RE: failSAFE IP - Looks good! Who's using it? (and why/why-not)R Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB70CC3E@tayexc19.americas.cpqcorp.net>   > -----Original Message-----= > From: Richard Maher [mailto:maher_rj@hotspamnotmail.com]=20 ! > Sent: November 11, 2005 2:23 AM  > To: Info-VAX@Mvb.Saic.Com @ > Subject: Re: failSAFE IP - Looks good! Who's using it? (and=20 > why/why-not) >=20 > Hi David,  >=20= > > See the description in the documentation, re: cautions=20  > about the minimum G > > number of interfaces in the scenario to prevent "false fail-overs".  >=20B > I searched the VMS Doc site for failSAFE and came up with the=20 > following @ > Technical Journal article. (Save anyone else from having to=20
 > look it up)  >=20@ > http://h71000.www7.hp.com/openvms/journal/v2/articles/tcpip.ht > ml#_Avoiding_Phantom_Failures  >=20@ > It doesn't look like that difficult a problem to solve, and=20 > along with the< > cluster failure split-brain problem it gives the System=20 > Managers something% > to plan for and do for a change :-)  >=207 > > > Just keep plugging away at 1.2.3.4 and it will=20  > automagically be sorted  > out!$ > > > What could be wrong with that? > > ; > > Services that aren't smart/persistent enough to keep=20  > hacking away at anD > > apparently dead address until it "magically" comes back to life. >=20@ > So what's Plan B? Fall over in a screaming heap? How do the=20 > options of PlansA > A and B vary in anyway with regard to failSAFE IP as opposed=20 
 > to say, the  > flushing of DNS cache? >=20B > Sure if somethin's not there, you can go off and cry about it=20 > or you canF > tell the network/system managers that's something's not right, while@ > continuing to get everybody out of the car and then putting=20 > them back inH > again, after waiting a respectible "pause-for-thought" interval beforeA > retring. I like to call mine T3$TIP_UNRESOLVED_SEARCH_DETENT=20  > (which defaults  > to 3 seconds)  >=20A > Apart from sulking, what other strategies are there? Polling=20  > is anathema toB > me! But without any meaningful events in this Unixie TCP/IP C=20 > world it's- > hard for a bloke to remain event driven :-(  >=20 > Regards Richard Maher  >=20@ > PS. Kerry, I know you have something to say about this (and=20
 > I'm sure we . > haven't fallen out) so can I please hear it? >=20  H Well, I have not looked at this for awhile, but as I recall, if one wereB looking at a high availability environment, you would likely use aG combination of the features described in the article mentioned earlier.   H DNS Alias with Load Broker provides system availability features and theH IP Failover provides capabilities to increase the link  availability andD transmit throughput. The Load broker actually uses host based serverE busy data to determine which nodes should get the new connections, so ' this is a better way of load balancing.   @ The other solutions are HW load balancers but they typically useG connection counting, DNS round robin etc to distribute the connections, H so they typically do not take actual server busy data to determine whichD servers are the most appropriate. And of course round robin does not; take into account node failures or how busy each serevr is.   E As a fyi, while I know zero details about this, a network ISV told me F that the following SASP (Server/Application State Protocol) IETF draftC is something they are looking at for their load balancing products:   < http://www.ietf.org/internet-drafts/draft-bivens-sasp-02.txt   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sun, 13 Nov 2005 12:59:22 +0800 3 From: "Richard Maher" <maher_rj@hotspamnotmail.com> H Subject: Re: failSAFE IP - Looks good! Who's using it? (and why/why-not)1 Message-ID: <dl6h9i$rap$1@news-02.connect.com.au>   	 Hi Kerry,   L Thanks very much for the reply!  (and please forgive my pouting in the other thread)   J > Well, I have not looked at this for awhile, but as I recall, if one wereD > looking at a high availability environment, you would likely use aI > combination of the features described in the article mentioned earlier.    Me too.   J > DNS Alias with Load Broker provides system availability features and theJ > IP Failover provides capabilities to increase the link  availability andF > transmit throughput. The Load broker actually uses host based serverG > busy data to determine which nodes should get the new connections, so ) > this is a better way of load balancing.   K I agree 100%. The requirement I have is that each application has two pipes L (network connections) and I'm happy to let DNS and Load Broker pick the nodeL best suited to serve the application's requirements but once that decision'sI been made then I *must* (for DECdtm transaction integrity reasons) ensure L that the second pipe arrives on the same cluster node. (Could be a differentK interface). The easiest way to achieve this would be to have the first pipe K return the UCX$INET_HOSTADDR of the Load-Broker-chosen cluster node so that L both connections end up in the same place. Why failSAFE IP is useful is thatL txn recovery (commit/abort) can be performed by *ANY* node in the cluster soK the ability to shift the firm-coded IP address means you don't have to wait 2 for a specific SCSNODE to come back up. I love it!  
 Thanks again.    Cheers Richard Maher  2 "Main, Kerry" <Kerry.Main@hp.com> wrote in messageL news:FD827B33AB0D9C4E92EACEEFEE2BA2FB70CC3E@tayexc19.americas.cpqcorp.net...   > -----Original Message-----: > From: Richard Maher [mailto:maher_rj@hotspamnotmail.com]! > Sent: November 11, 2005 2:23 AM  > To: Info-VAX@Mvb.Saic.Com = > Subject: Re: failSAFE IP - Looks good! Who's using it? (and  > why/why-not) >  > Hi David,  > : > > See the description in the documentation, re: cautions > about the minimum G > > number of interfaces in the scenario to prevent "false fail-overs".  > ? > I searched the VMS Doc site for failSAFE and came up with the  > following = > Technical Journal article. (Save anyone else from having to 
 > look it up)  > @ > http://h71000.www7.hp.com/openvms/journal/v2/articles/tcpip.ht > ml#_Avoiding_Phantom_Failures  > = > It doesn't look like that difficult a problem to solve, and  > along with the9 > cluster failure split-brain problem it gives the System  > Managers something% > to plan for and do for a change :-)  > 4 > > > Just keep plugging away at 1.2.3.4 and it will > automagically be sorted  > out!$ > > > What could be wrong with that? > > 8 > > Services that aren't smart/persistent enough to keep > hacking away at anD > > apparently dead address until it "magically" comes back to life. > = > So what's Plan B? Fall over in a screaming heap? How do the  > options of Plans> > A and B vary in anyway with regard to failSAFE IP as opposed
 > to say, the  > flushing of DNS cache? > ? > Sure if somethin's not there, you can go off and cry about it  > or you canF > tell the network/system managers that's something's not right, while= > continuing to get everybody out of the car and then putting  > them back inH > again, after waiting a respectible "pause-for-thought" interval before> > retring. I like to call mine T3$TIP_UNRESOLVED_SEARCH_DETENT > (which defaults  > to 3 seconds)  > > > Apart from sulking, what other strategies are there? Polling > is anathema to? > me! But without any meaningful events in this Unixie TCP/IP C  > world it's- > hard for a bloke to remain event driven :-(  >  > Regards Richard Maher  > = > PS. Kerry, I know you have something to say about this (and 
 > I'm sure we . > haven't fallen out) so can I please hear it? >   H Well, I have not looked at this for awhile, but as I recall, if one wereB looking at a high availability environment, you would likely use aG combination of the features described in the article mentioned earlier.   H DNS Alias with Load Broker provides system availability features and theH IP Failover provides capabilities to increase the link  availability andD transmit throughput. The Load broker actually uses host based serverE busy data to determine which nodes should get the new connections, so ' this is a better way of load balancing.   @ The other solutions are HW load balancers but they typically useG connection counting, DNS round robin etc to distribute the connections, H so they typically do not take actual server busy data to determine whichD servers are the most appropriate. And of course round robin does not; take into account node failures or how busy each serevr is.   E As a fyi, while I know zero details about this, a network ISV told me F that the following SASP (Server/Application State Protocol) IETF draftC is something they are looking at for their load balancing products:   < http://www.ietf.org/internet-drafts/draft-bivens-sasp-02.txt   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)   4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sun, 13 Nov 2005 00:59:09 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> H Subject: Re: failSAFE IP - Looks good! Who's using it? (and why/why-not), Message-ID: <4376D60F.CB76854C@teksavvy.com>   "Main, Kerry" wrote:J > DNS Alias with Load Broker provides system availability features and theJ > IP Failover provides capabilities to increase the link  availability andF > transmit throughput. The Load broker actually uses host based serverG > busy data to determine which nodes should get the new connections, so ) > this is a better way of load balancing.     H Out of curiosity, if load balancing is done via DNS tricks, what sort of& TTL values are used for DNS entries ?   D For instance, if you try to download a standard MS Frontage designedH page, you will have a gazillion HTTP transactions. Once your client getsH the IP for WWW.CHOCOLATE.COM, it will reused that IP for the duration ofF the transaction and if you skip to the next page, it will re-use it asD well, unless you have an abnormally low TTL of just a second or two.   ------------------------------    Date: 12 Nov 2005 12:35:10 -0800< From: "Hein RMS van den Heuvel" <heinvandenheuvel@gmail.com>P Subject: Re: Lack of response on c.o.v. (was:Re: File spec wildcard match test?)B Message-ID: <1131827710.159377.31500@g43g2000cwa.googlegroups.com>  G IMHO the HP ITRC  has become the more likely place to get technical VMS  questions answerred 'for free'.   D http://forums1.itrc.hp.com/service/forums/familyhome.do?familyId=288   fwiw,  Hein.    ------------------------------  # Date: Sat, 12 Nov 2005 22:01:03 GMT  From: "Jim" <j.n@nospam.com>P Subject: Re: Lack of response on c.o.v. (was:Re: File spec wildcard match test?)> Message-ID: <zCtdf.18717$dO2.12413@newssvr29.news.prodigy.net>  0 "AEF" <spamsink2001@yahoo.com> wrote in message < news:1131818226.446880.25040@z14g2000cwz.googlegroups.com... > Jim wrote:5 >> "BRAD" <bradhamilton@comcast.net> wrote in message ' >> news:437541DD.8050607@comcast.net...  >> > AEF wrote:  >> > <snip> L >> >> I hear you! I have gotten the full range of understanding to my posts.C >> >> Sometimes I get exactly the answer I need. Other times people J >> >> completely misunderstand me and post stuff that is of no help. Still  >> >> other times I get nothing. >> >> H >> >> E.g., I recently posted a question about why I can't get my DECnetJ >> >> configurators to work. No response at all. Is it really true that noK >> >> one who reads cov can answer this question? Have all the configurator K >> >> experts put me in their kill file? Oh well. Maybe the configurator is G >> >> just a joke: It doesn't really work, but simply pads the manual a H >> >> little. ;-) (Just kidding.) Maybe I'll try ITRC or Ask the Wizard. >> >K >> > Perhaps not many folks are left who know the answer to your question - J >> > after all, it _is_ about DECnet ("Phase IV", as well!); by the time IH >> > started working as a System Mangler in the late 1990s, the network 
 >> > staffJ >> > at my company had no clue what "decknet" was, and supported IP only.  >> > To H >> > save ourselves much heartache, my colleagues and I spent much time  >> > making K >> > sure the application needed as little use of "decknet" as possible.  I H >> > decided that learning the ins and outs of Phase IV (and V) was not 
 >> > worth >> > it to me (or my job). >> >H >> > I appreciate your feelings, including the rant that follows; I justJ >> > decided that it wasn't worth the time or energy tilting at windmills. >> >- >> > "Forget it, Jake... It's Chinatown." :-)  >> > >> > <snip> G >> Sorry to say, I didn't see the earlier post.  The OP should specify  
 >> whetherK >> he is trying to get Decnet Phase IV or Decnet/OSI going.  These are two   >> very  >> different beasts. >  > DECnet phase IV. See > w > http://groups.google.com/group/comp.os.vms/browse_frm/thread/ad73ad2d5b26956b/f88d636370a39c67?hl=en#f88d636370a39c67  > L The problem lies at the remote node.  Those other commands you cited excute : on the local node and you have the right to execute there.L If I remember correctly, the configurator module executes on a remote node, G and all of the accounts and passwords must be correct for execution to  	 complete. B The DECnet Phase IV manual should describe how this command works.9 Exactly why does the account Mystery_User come into play?  Jim    ------------------------------  % Date: Sat, 12 Nov 2005 11:53:15 -0800 1 From: Chuck Forsberg WA7KGX N2469R <caf@omen.com> $ Subject: Re: Microvax image to simh?5 Message-ID: <pan.2005.11.12.19.53.13.461687@omen.com>   5 On Fri, 11 Nov 2005 19:34:05 -0800, Alan Greig wrote:  > J > Backup/image the boot disk of your MVII to a backup saveset. Install VMSI > and Boot simh using the documented procedures (or just boot stand alone K > backup) and restore the image to another virtual disk. Boot simh from the  > oher disk 5 I don't have a VMS image (w/compiler) to put on simh. * > Good to see you are stiill around Chuck. Me too? >> Developer of Industrial ZMODEM(Tm) for Embedded Applications = >>   Omen Technology Inc      "The High Reliability Software" ? >> 10255 NW Old Cornelius Pass Portland OR 97231   FAX 629-0665    --  < Chuck Forsberg    caf@omen.com   www.omen.com   503-614-0430< Developer of Industrial ZMODEM(Tm) for Embedded Applications:   Omen Technology Inc      "The High Reliability Software"< 10255 NW Old Cornelius Pass Portland OR 97231   FAX 629-0665   ------------------------------  % Date: Sat, 12 Nov 2005 20:52:58 -0500 ( From: Wayne Sewell <wayne@tachysoft.com>6 Subject: re: Request for feedback - BACKUP enhancement/ Message-ID: <00A4CB53.86A3A3EB.7@tachysoft.com>   8 >From: "Guy Peleg" <guy.peleg@remove_this_header@hp.com> >X-Newsgroups: comp.os.vms7 >Subject: Re: Request for feedback - BACKUP enhancement & >Date: Thu, 10 Nov 2005 08:00:30 +0200    3 >I would like to thank everyone for their feedback.  > 5 >While I did not response to every one of the replies 7 >(or the 100 extra messages sent directly to me) I sure  >read it all.....  > : >We have some research to do, although it looks like there: >is an interest in these features (while implementation is( >yet TBD) and I'll keep you guys posted. > 2 >Here is a quick update on latests happenings with >BACKUP: >    <changes deleted>    > # >Again thank you for your feedback.  >     K While you are making improvements, could anything possibly be done with the M API?  AFAIK, the interface has not changed in the slightest since the API was L first introduced in 7.1 or 7.2 or whichever it was.  Don't get me wrong, theM backup API was a great enhancement and I use the hell out of it, but there is K more that could be done to make it extremely powerful for customized backup  operations.   N As probably the heaviest user of the api on earth, I have asked for this stuff9 a couple of times ready, but I thought I would try again.    There are two basic areas:    > 1.  more callbacks and greater control of the backup operation& 2.  help with parsing the command line  
 Callbacks:  K Nearly all of the existing callbacks are passive.  Basically backup lets me M know when it does things, possibly with useful information, but there is very G little oportunity to change its behavior once the backup has started.     N For instance, I would like to be able to specify the label that will go on theN tape in real time, on the first and especially subsequent volumes.  To do thatI now, I have to do system service and runtime library intercepts, a rather  fragile mechanism.  M It would be awesome if there was a label generation callback.  When backup is M in the process of deriving a label from the saveset name, use the callback to O see if I want to provide you with a label.  If I do, forget the generated label & and write what I give you on the tape.  J This would eliminate 90% of the special purpose code in the tapesys backupM module and would allow me to use established/supported interfaces for greater 0 reliability and immunity to vms version changes.     The backup command line:    O The backup command is fairly complex and has a lot of qualifiers.  The api does I not deal with the command line at all and expects all of the values to be L already parsed and placed into a fixed length parameter list, similar to theL item lists used for various system services.  I guess the assumption is thatL all custom backup programs will generate this list dynamically from internalH values.  In truth I have done this in simple programs used for testing.   J However, an equally likely scenario is that the custom backup program is aM wrapper around regular vms backup.  This means that all backup parameters and N qualifiers are identical to those used by straight backup, even if the wrapperF doesn't do anything with them.  So basically the same parsing is done,L resulting in an identical parameter list passed to backup$start in the api. O Unfortunately, the code to do this is in backup.exe, *not* backupshr.  So it is L not available to someone using the api.  All of the command line parsing andN parameter list building code has to be replicated from scratch, even though itL is logically identical to what is in backup.exe.  Worse, you have to monitorM new versions of backup for new qualifiers and modify your code to add them to F the list.  If you forget to do it, the customers complain that the new qualifiers are being ignored.   @ It would be a great service to the api user to simply *move* theL parameter/qualifier processing code from backup.exe to backupshr and make itO available as a callable procedure (parse_backup_command_line or some such).  No O input is required; the routine can simply use the cli routines to check all the K parameters/qualifiers as it always does.  The output would be the parameter  list passed to backup$start.    N So the custom backup program would call parse_backup_command_line, massage the: generated parameter list if needed, and call backup$start.  N backup.exe would do basically the same thing.  IIRC from examining the source,J the parsing code is a called procedure already, so this would be a case ofG simply moving that procedure from one module to another and giving it a N transfer vector in the shareable image.  And adding more error checking for us0 irrational, unpredictable users, of course.  :-)   Wayne O =============================================================================== N Wayne Sewell, Tachyon Software Consulting  (281)812-0738   wayne@tachysoft.com; http://www.tachysoft.com/www/tachyon.html and wayne.html    O =============================================================================== P Jake Blues:"You traded the Caddy for a microphone? ...... Okay, I can buy that."   ------------------------------  % Date: Sat, 12 Nov 2005 21:20:15 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 6 Subject: Re: Request for feedback - BACKUP enhancement, Message-ID: <4376A2CF.D71C61B9@teksavvy.com>   Wayne Sewell wrote: O > It would be awesome if there was a label generation callback.  When backup is O > in the process of deriving a label from the saveset name, use the callback to Q > see if I want to provide you with a label.  If I do, forget the generated label ( > and write what I give you on the tape.    @ I agree 100%.  In fact, even BACKUP.EXE should have better labelG handling.  If you are not sure how many tapes your backup will require, E you need to still allocate a whole bunch of tapes and specify them in B the backup command and after the backup has completed, you need to: search the log file to see which tapes were actually used.  F In a tape library with pre-initialised tapes, you don't want backup to change tape labels on you.   ------------------------------  % Date: Saturday, 12 NOV 2005 23:46 EST  From: bjj101@arlvax.arl.psu.edu 6 Subject: Re: Request for feedback - BACKUP enhancement. Message-ID: <dl6in7$1av4$1@f04n12.cac.psu.edu>   >Tape striping  = Looks to me like current tape drives are hitting 100+ MB/sec. > I find it hard to believe any VMS system can keep 1 such drive spinning much less 2.   A In particular, backup's inability to do simultaneous read & write 6 (as mentioned by the Italians) has been a real killer.   ------------------------------  % Date: Sun, 13 Nov 2005 10:41:18 +0800 3 From: "Richard Maher" <maher_rj@hotspamnotmail.com> . Subject: Speaking of "fulcrumizing the foible"1 Message-ID: <dl696m$dmc$1@news-02.connect.com.au>    Hi,     Just read this in openvms.org: -  9 http://www.openvms.org/stories.php?story=05/11/09/0139151   I It's about Oracle 9iRAC in a long distance disater-tolerant cluster test. ) Well worth a look if you haven't seen it.   9 Does anyone know if they had Cache-Fusion up and running?    Regards Richard Maher   I PS. It's great to see a popular database making the most of VMS Clusters! # Just a shame it's the wrong one :-(    ------------------------------  % Date: Sat, 12 Nov 2005 16:03:05 -0500 , From: "Richard Tomkins" <tomkinsr@istop.com>. Subject: Re: Strange Network performance issue7 Message-ID: <1131829545_1301@spool6-east.superfeed.net>   J Some systems will run at Full Duplex, some will only run at Half Duplex. IJ have had similar problems like this in the past. With some older NIC cardsI we used to produce, they did not auto-negotiate Speed or Duplex  properly J and some of them even worked for a while and then went back to negotiationI during use, knocking everything off the network. You can only detect this L type of fault with a proper network analysis tool, or assume it is happening/ if DECnet is up and down with great regularity.   K You said you were using a HUB. Don't forget, a HUB repeats all the Ethernet K traffic on all ports at the same time. A SWITCH on the other hand has three G advantages that immediately come to mind. !. They only pass the traffic J between the two communicating nodes and not all the ports as they actuallyJ learn the route. 2. They re-manufacture / re-time the Ethernet signals. 3.L They enable 100MB/s and 10MB/s interfaces to live on individual ports and doG not force all ports to the lowest common denominator for speed, 10MB/s.    rtt       5 "Jeff Cameron" <roktsci@comcast.net> wrote in message * news:BF9A95C6.17B7D%roktsci@comcast.net...G > I'm having a very Strange Network issue concerning performance that I  would ; > like to solicit comments and suggestions from all of you.  > 2 > I have three separate VMS systems outlined here: >  > Node: MAUI > VAXStation 3100 	 > VMS 5.6  > 10Base-T Ethernet FDX  >  > Node: PELE > Alpha DS20 > VMS 7.3-2  > 100Base-T Ethernet FDX > 
 > Node: KUPUA  > Alpha 2100 > VMS 7.3-1  > 100Base-T Ethernet FDX > 5 > All nodes are running Decnet Phase IV and Multinet.  > L > Nodes are connected through a single common Hub and all reside on the sameH > IP subnet. All hub ports have Auto-Negotiate disabled, and Full Duplex set.2 > They also all share the same DECNet Area number. > I > The hub makes a connection to the company LAN. The idea was to keep the 6 > Decnet and Intra-VMS IP traffic off the Company LAN. > G > Decnet Copy and FTP Speeds are consistent with one another, and shown  here:  > ! > MAUI <-> PELE  : 250K Bytes/Sec ! > MAUI <-> KUPUA : 250K Bytes/Sec ! > PELE <-> KUPUA :  17K Bytes/Sec  > D > FTP speeds from PCs located on different IP subnet on company LAN: > ! > PCs  <-> MAUI  : 242K Bytes/Sec ! > PCs  <-> PELE  : 247K Bytes/Sec ! > PCs  <-> KUPUA : 245K Bytes/Sec  > D > The question is why is the throughput between the two Alpha's with	 100Base-T " > so abysmally slow, DECNet or IP? >  > Any ideas? >  > Thank you in advance.  > Jeff Cameron >  >       Q ----== Posted via Newsfeeds.Com - Unlimited-Unrestricted-Secure Usenet News==---- S http://www.newsfeeds.com The #1 Newsgroup Service in the World! 120,000+ Newsgroups K ----= East and West-Coast Server Farms - Total Privacy via Encryption =----    ------------------------------    Date: 12 Nov 2005 12:46:01 -0800$ From: "AEF" <spamsink2001@yahoo.com>: Subject: Re: Taking backup of disks of production system..B Message-ID: <1131828361.224356.68370@g43g2000cwa.googlegroups.com>   MUSTAFA ATAKAN wrote:  > Hi,  > I > Our cluster system is 7x24 online and there is always something reading D > or writing to disks. We are taking full image backup on Sunday andG > incremental backups on other days of a week without shutting down any I > program on the system. In the log files of the backup, there are always G > a warning messages saying "X file is open for write by another user",  > but still backuping the file.   C It sounds like you are using BACKUP/IMAGE/IGNORE=INTERLOCK. Can you  confirm?  G > My question is, suppose that we are using 20 disks on our storage and H > all of them are gone in a disaster. However, I have got the image tapeJ > backups of the disks (with the above method) and wanted to restore theseH > backups to another 20 disks-storage device (suppose that it is also OKH > to lose some file that is produced or changed after backup). But, I amH > not sure whether it will work or not. Because, when looking at the log3 > file of the backups, there are always saying that   E Well, to be sure, do a BACKUP/LIST of one of the tapes and see if the D files you got warnings about in your log file are present. If you're6 using /IGNORE=INTERLOCK, they'll most likely be there.  A Now, whether the restore will be okay depends on what writing was C occuring during the backups. Only you can answer that. Depending on C what's being written, if anything, during your backups, it could be E fine or it could result in inconsitent files on the tape. You need to B find out what your apps and/or users are doing during the backups.   > A >  "diskX:[000000]QUOTA.SYS;1 is open for write by another user".   D In the future, please post the COMPLETE error message, including the       %FACILITY-S-IDENT,  
 part. thanks.   I > If some of files gave this kind of warning when taking the image backup F > of a disk, is it possible to restore this backup to another disk andI > still continue to work? Or is there another way to take the snapshot of 3 > the disk and then taking backup of this snapshot?   A Well, this particular file, I think, isn't of much concern if you G rebuild your disks after the restore. The quotas should be recalculated E during the rebuild, IIRC (I haven't used disk quotas in a long time). G But other files could easily be a problem. Again, it all depends on the 4 particulars of your system usage during the backups.  D The best thing, of course, is to backup disks with no files open forC write. There are other ways, but that depends on the particulars of 
 your site.  " I think I'm repeating myself here!  E Basically, without more information from you about the particulars of ? your site, we can't really be of further help (well, I can't!).   E > Note: My OS version OpenVMS 7-3-2 and my disks are on HSG80 storage    AEF    ------------------------------  % Date: Sun, 13 Nov 2005 00:14:13 +0200 0 From: "MUSTAFA ATAKAN" <matakan@inteltek.com.tr>: Subject: RE: Taking backup of disks of production system..L Message-ID: <F014DACB8BE63442993543B780A2F01801AD7486@asteriks.inteltek.ist>   =20    > -----Original Message-----. > From: AEF [mailto:spamsink2001@yahoo.com]=20, > Sent: Saturday, November 12, 2005 10:46 PM > To: Info-VAX@Mvb.Saic.Com < > Subject: Re: Taking backup of disks of production system.. >=20 >=20 > MUSTAFA ATAKAN wrote:  > > Hi,  > > F > > Our cluster system is 7x24 online and there is always something=20< > > reading or writing to disks. We are taking full image=20 > backup on Sunday=20 > > > and incremental backups on other days of a week without=20 > shutting down=20A > > any program on the system. In the log files of the backup,=20  > there are=20A > > always a warning messages saying "X file is open for write=20  > by another=20 ( > > user", but still backuping the file. >=20B > It sounds like you are using BACKUP/IMAGE/IGNORE=3DINTERLOCK.=20 > Can you confirm?   Yes...   >=20@ > > My question is, suppose that we are using 20 disks on our=20 > storage and=20B > > all of them are gone in a disaster. However, I have got the=20 > image tape=20 I > > backups of the disks (with the above method) and wanted to restore=20 @ > > these backups to another 20 disks-storage device (suppose=20 > that it is=20 B > > also OK to lose some file that is produced or changed after=20
 > backup).=20 ? > > But, I am not sure whether it will work or not. Because,=20  > when looking=20 @ > > at the log file of the backups, there are always saying that >=20? > Well, to be sure, do a BACKUP/LIST of one of the tapes and=20 A > see if the files you got warnings about in your log file are=20 ? > present. If you're using /IGNORE=3DINTERLOCK, they'll most=20  > likely be there. >=20B > Now, whether the restore will be okay depends on what writing=20? > was occuring during the backups. Only you can answer that.=20 @ > Depending on what's being written, if anything, during your=20@ > backups, it could be fine or it could result in inconsitent=20B > files on the tape. You need to find out what your apps and/or=20% > users are doing during the backups.   G As far as i undestand, you say that if the file in the backup save set, E there is no worry whether it is open or not during backup. But in the $ help page of backup, it is said that  6 Use of the INTERLOCK keyword overrides these file data?         integrity interlocks. The data that BACKUP subsequently A         transfers can then contain corrupted data for open files. D         Also, all cases in which these data corruptions can occur inC         the data that BACKUP transfers are not be reliably reported D         to you; in other words, silent data corruptions are possible$         within the transferred data.  D So, altough the backup utility writes the open file to the save-set,/ there is still possibilty that it is corrupted.    >=20 > > C > >  "diskX:[000000]QUOTA.SYS;1 is open for write by another user".  >=20F > In the future, please post the COMPLETE error message, including the >=20 >     %FACILITY-S-IDENT, >=20 > part. thanks.  ..E %BACKUP-W-ACCONFLICT, $diskX:[000000]QUOTA.SYS;1 is open for write by  another user3 %BACKUP-S-COPIED, copied $diskX:[000000]QUOTA.SYS;1  ..   >=20G > > If some of files gave this kind of warning when taking the image=20 I > > backup of a disk, is it possible to restore this backup to another=20 B > > disk and still continue to work? Or is there another way to=20
 > take the=20 A > > snapshot of the disk and then taking backup of this snapshot?  >=20B > Well, this particular file, I think, isn't of much concern if=20@ > you rebuild your disks after the restore. The quotas should=20B > be recalculated during the rebuild, IIRC (I haven't used disk=20 > quotas in a long time). = > But other files could easily be a problem. Again, it all=20 E > depends on the particulars of your system usage during the backups.  >=20@ > The best thing, of course, is to backup disks with no files=20B > open for write. There are other ways, but that depends on the=20 > particulars of your site.   E Yes, i am wondering these ways you recommend... What else is required D for a suggestion? My only requirement is that i can not stop the theD processes during backup (since it is 7x24) and i want to restore theE data on new empty disks without any corruption. Suppose that i am not G worrying about data loss (that is modified or created after backup).=20    >=20$ > I think I'm repeating myself here! >=20; > Basically, without more information from you about the=20 A > particulars of your site, we can't really be of further help=20  > (well, I can't!).  >=20G > > Note: My OS version OpenVMS 7-3-2 and my disks are on HSG80 storage  >=20 > AEF  >=20 >=20   ------------------------------  # Date: Sat, 12 Nov 2005 22:11:32 GMT  From: "Jim" <j.n@nospam.com>: Subject: Re: Taking backup of disks of production system..= Message-ID: <oMtdf.18719$dO2.7742@newssvr29.news.prodigy.net>   L Yes, VMS Backup will keep going because that message is purely informative, M It does indicate that the backup of that particular file will be incomplete.  I To surpress the message, specify /Ignore=Interlock on the backup command.   L You most assuredly will be able to restore a disk from backup saveset under I these circumstances, but those files which are open will not be restored  K completely.  As for taking a snapshot, if you can't do it with VMS Backup,  # you can't do it with anything else.  Jim < "MUSTAFA ATAKAN" <matakan@inteltek.com.tr> wrote in message F news:F014DACB8BE63442993543B780A2F01801AD742B@asteriks.inteltek.ist... Hi,   G Our cluster system is 7x24 online and there is always something reading B or writing to disks. We are taking full image backup on Sunday andE incremental backups on other days of a week without shutting down any G program on the system. In the log files of the backup, there are always E a warning messages saying "X file is open for write by another user",  but still backuping the file.   E My question is, suppose that we are using 20 disks on our storage and F all of them are gone in a disaster. However, I have got the image tapeH backups of the disks (with the above method) and wanted to restore theseF backups to another 20 disks-storage device (suppose that it is also OKF to lose some file that is produced or changed after backup). But, I amF not sure whether it will work or not. Because, when looking at the log1 file of the backups, there are always saying that   ?  "diskX:[000000]QUOTA.SYS;1 is open for write by another user".   G If some of files gave this kind of warning when taking the image backup D of a disk, is it possible to restore this backup to another disk andG still continue to work? Or is there another way to take the snapshot of 1 the disk and then taking backup of this snapshot?     D Note: My OS version OpenVMS 7-3-2 and my disks are on HSG80 storage    ------------------------------  % Date: Sat, 12 Nov 2005 18:10:23 -0500 ' From: Dave Froble <davef@tsoft-inc.com> : Subject: Re: Taking backup of disks of production system..0 Message-ID: <11nctihq29igm7b@corp.supernews.com>   MUSTAFA ATAKAN wrote:  > Hi,  > I > Our cluster system is 7x24 online and there is always something reading D > or writing to disks. We are taking full image backup on Sunday andG > incremental backups on other days of a week without shutting down any I > program on the system. In the log files of the backup, there are always G > a warning messages saying "X file is open for write by another user",   > but still backuping the file.  > G > My question is, suppose that we are using 20 disks on our storage and H > all of them are gone in a disaster. However, I have got the image tapeJ > backups of the disks (with the above method) and wanted to restore theseH > backups to another 20 disks-storage device (suppose that it is also OKH > to lose some file that is produced or changed after backup). But, I amH > not sure whether it will work or not. Because, when looking at the log4 > file of the backups, there are always saying that  > B >  "diskX:[000000]QUOTA.SYS;1 is open for write by another user".  > I > If some of files gave this kind of warning when taking the image backup F > of a disk, is it possible to restore this backup to another disk andI > still continue to work? Or is there another way to take the snapshot of 3 > the disk and then taking backup of this snapshot?  >  > F > Note: My OS version OpenVMS 7-3-2 and my disks are on HSG80 storage   E Based upon your specifications, ie; you can live with data not being  # concurrent, then the answer is yes.   E You're using /IGNORE=INTERLOCK, so the file will be copied, however,  C BACKUP gives you a warning that the file is open, and could change  I before, during, and after the file is copied by BACKUP.  Regardless, all  J of the file, as it exists during the copy, will be in the backup save set.  G With 20 volumns, if there is activity, WRITE activity, and data for an  I application is spread over multiple volumes, or even on the same volume,  9 there is no guarantee that transactions will be complete.   H For example, a CUSTOMER file and an AR ITEM file.  An ITEM is added, or G modified, thus changing the customer's balance, which for this purpose  F we'll say is in the customer file.  The CUSTOMER file is copied, then I the transaction occurs, then the ITEM file is copied.  The data in the 2  B files will be out of sync, should you need to restore from backup.  D You state that you can live with this.  If so, then you're Ok doing E backups as you state above.  Note that the majority would not accept   this as a valid backup.    --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Sat, 12 Nov 2005 16:53:12 -0700 4 From: Norman Lastovica <norman.lastovica@oracle.com>: Subject: Re: Taking backup of disks of production system..* Message-ID: <43768068.22C04C04@oracle.com>  7 I would be tempted to suggest an extra step or two that 7 might help increase the chance of being able to restore  good data and valid files...  : - just prior to the backup run, for each indexed file that= is commonly open for write access (including SYSUAF, MAILAUF, G RIGHTSLIST, etc), do "$ CONVERT/SHARE <file> <backupfile> /FAST/NOSORT" = this will provide you with an internally consistant file that : won't possibily be corrupt should the normal backup manage< to copy the file such that there is a partial update (ie, an; update to blocks X and Y is in progress and the backup gets : the old copy of X and the new copy of Y - perhaps an index3 split or something along these lines was going on).   @ - make certain that all of your backups do /CHECKSUM alwys.  any; backup worth doing is worth doing "right".  adding checksum > protection costs little and adds another layer of security for	 the data.   > - do /COMPARE to make sure that your output matches the input.@ this can also "pre-detect" cases where updates to files happened? during the backup and you can use this as an indicator of files = that might require some additional attention as per the first  bullet.    Dave Froble wrote: >  > MUSTAFA ATAKAN wrote:  > > Hi,  > > K > > Our cluster system is 7x24 online and there is always something reading F > > or writing to disks. We are taking full image backup on Sunday andI > > incremental backups on other days of a week without shutting down any K > > program on the system. In the log files of the backup, there are always I > > a warning messages saying "X file is open for write by another user", ! > > but still backuping the file.  > > I > > My question is, suppose that we are using 20 disks on our storage and J > > all of them are gone in a disaster. However, I have got the image tapeL > > backups of the disks (with the above method) and wanted to restore theseJ > > backups to another 20 disks-storage device (suppose that it is also OKJ > > to lose some file that is produced or changed after backup). But, I amJ > > not sure whether it will work or not. Because, when looking at the log5 > > file of the backups, there are always saying that  > > C > >  "diskX:[000000]QUOTA.SYS;1 is open for write by another user".  > > K > > If some of files gave this kind of warning when taking the image backup H > > of a disk, is it possible to restore this backup to another disk andK > > still continue to work? Or is there another way to take the snapshot of 5 > > the disk and then taking backup of this snapshot?  > >  > > G > > Note: My OS version OpenVMS 7-3-2 and my disks are on HSG80 storage  > F > Based upon your specifications, ie; you can live with data not being% > concurrent, then the answer is yes.  > F > You're using /IGNORE=INTERLOCK, so the file will be copied, however,D > BACKUP gives you a warning that the file is open, and could changeJ > before, during, and after the file is copied by BACKUP.  Regardless, allL > of the file, as it exists during the copy, will be in the backup save set. > H > With 20 volumns, if there is activity, WRITE activity, and data for anJ > application is spread over multiple volumes, or even on the same volume,; > there is no guarantee that transactions will be complete.  > I > For example, a CUSTOMER file and an AR ITEM file.  An ITEM is added, or H > modified, thus changing the customer's balance, which for this purposeG > we'll say is in the customer file.  The CUSTOMER file is copied, then J > the transaction occurs, then the ITEM file is copied.  The data in the 2D > files will be out of sync, should you need to restore from backup. > E > You state that you can live with this.  If so, then you're Ok doing F > backups as you state above.  Note that the majority would not accept > this as a valid backup.  >  > --6 > David Froble                       Tel: 724-529-04506 > Dave Froble Enterprises, Inc.      Fax: 724-529-0596@ > DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com > 170 Grimplin Road  > Vanderbilt, PA  15486    --  	 - - - - - 0  opinions expressed here are mine and mine alone.  and certainly are not intended in any way to 0  express or represent any opinions or commitment  of oracle corporation.   *  norman lastovica / oracle rdb engineering   ------------------------------    Date: 12 Nov 2005 16:46:23 -0800< From: "Hein RMS van den Heuvel" <heinvandenheuvel@gmail.com>: Subject: Re: Taking backup of disks of production system..B Message-ID: <1131842783.766731.89760@z14g2000cwz.googlegroups.com>   fyi...  B Mustafa also posted this question in the itrc forum where I gave a reply.K http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=974326    Hein.    ------------------------------    Date: 12 Nov 2005 19:55:12 -0100* From: "Michael Kraemer" <M.Kraemer@gsi.de>F Subject: Re: Video card Digital PCI ZLXp-L1 compatibility with OpenVMS0 Message-ID: <4376489F.MD-1.4.4.M.Kraemer@gsi.de>   > >  > >what about the ZLXp-L2 ?  > > > The same. See http://www.compaq.com/info/SP4508/SP4508PF.PDF > G > As FredK more than once stated, drivers previously in OPEN3D for this K > 3D cards are now at best in 2D mode in VMS. No more OPEN3D. Deinstall it. K > It won't work with VMS V7.3-2/MOTIF V1.3-1 and up (it will sometimes hang E > the X11 server in a loop - at prio 6 - cause of MOTIF API changes). C > Open3D V4.9B was the last (from 2001 for VMS V7.2-x up to V7.3-1)  > 9 > If you need 3D, then you need at least EV6 systems now.  >    now that's a bit confusing. 9 I don't need 3D, I just want a decent GUI (2D/Motif) on a = machine that once was sold as a workstation (with a ZLXp-L2). D VMS release notes tell me that I need Open3D to get *any* gfx out of it. A And indeed it seems: no Open3D, no gfx, same symptoms as the OP.  B Now you tell me *not* to use Open3D. Fine. How can I get gfx now ?   ------------------------------    Date: 12 Nov 2005 21:05:00 +01006 From: peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER)F Subject: Re: Video card Digital PCI ZLXp-L1 compatibility with OpenVMS, Message-ID: <437658fc$1@news.langstoeger.at>  ] In article <4376489F.MD-1.4.4.M.Kraemer@gsi.de>, "Michael Kraemer" <M.Kraemer@gsi.de> writes:  >now that's a bit confusing.: >I don't need 3D, I just want a decent GUI (2D/Motif) on a> >machine that once was sold as a workstation (with a ZLXp-L2).  O If it was sold with a ZLXp-L2 it was surely sold as a 'powerful 3D workstation'   J >VMS release notes tell me that I need Open3D to get *any* gfx out of it. B >And indeed it seems: no Open3D, no gfx, same symptoms as the OP. C >Now you tell me *not* to use Open3D. Fine. How can I get gfx now ?   D 1) Use a non-current VMS version (upto V7.3-1) with MOTIF and Open3D< and your Alpha/ZLXp-L2 (ZLXp-L2 requires Open3D even for 2D) orJ 2) Get another graphic adapter (ZLXp-E1, 3D30, 4D20, ...) which works with/ current VMS versions (in 2D mode) in your Alpha  orE 3) Get another graphic adapter (300, 350, VX1, ...) and another Alpha # system (EV6 and up) to stick it in.    --   Peter "EPLAN" LANGSTOEGER % Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------    Date: 12 Nov 2005 17:16:07 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) P Subject: Re: Will Rich Marcello either come clean or do some proper marketingand3 Message-ID: <I$QIfgcQtfsM@eisner.encompasserve.org>   \ In article <43767558.DDEB749C@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes:  E > The smart thing to do would be to get VMS to support all of the NSK  > functionality   G Like the requirement to run on specialized hardware much more expensive  than that used for VMS today ?   ------------------------------  % Date: Sat, 12 Nov 2005 18:31:27 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> P Subject: Re: Will Rich Marcello either come clean or do some proper marketingand, Message-ID: <43767B49.59B1A75D@teksavvy.com>   Larry Kilgallen wrote:I > Like the requirement to run on specialized hardware much more expensive   > than that used for VMS today ?    G VMS can run on workstations, it can run on wildfire class machines with / the galaxy interface between instances of VMS.    C VMS had fault tolerance, but that didn't prevent it from running on 7 platforms that didn't have the fault tolerant hardware.   H Supporting NSK features would allow the NSK apps to run on VMS, but theyG would still want to make falt tolerant hardware for those applications.    ------------------------------  % Date: Sat, 12 Nov 2005 22:50:49 -0500 ' From: "Main, Kerry" <Kerry.Main@hp.com> P Subject: RE: Will Rich Marcello either come clean or do some proper marketingandR Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB70CC3D@tayexc19.americas.cpqcorp.net>   > -----Original Message-----9 > From: JF Mezei [mailto:jfmezei.spamnot@teksavvy.com]=20 ! > Sent: November 12, 2005 6:31 PM  > To: Info-VAX@Mvb.Saic.Com A > Subject: Re: Will Rich Marcello either come clean or do some=20  > proper marketingand  >=20 > Larry Kilgallen wrote:? > > Like the requirement to run on specialized hardware much=20  > more expensive" > > than that used for VMS today ? >=20 >=20> > VMS can run on workstations, it can run on wildfire class=20 > machines with 3 > the galaxy interface between instances of VMS.=20  >=20E > VMS had fault tolerance, but that didn't prevent it from running on 9 > platforms that didn't have the fault tolerant hardware.  >=20? > Supporting NSK features would allow the NSK apps to run on=20  > VMS, but they > > would still want to make falt tolerant hardware for those=20 > applications.  >=20  E I guess that would depend on whether you thought the future for fault D tolerant computing was going to be based on hardware or software.=20  E The NSK folks have already stated they will be using standard Itanium C systems in a triplicated configuration that is essentially software % based to achieve its fault tolerance.   E Imho, there are just to many ways to achieve fault tolerant solutions H with software strategies than to design very high end expensive hardwareG solutions. In OpenVMS terms, that would include using technologies such @ as RTR which, similar to HW fault tolerance, ensures a committed transaction is never lost.   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sun, 13 Nov 2005 12:37:40 +0800 3 From: "Richard Maher" <maher_rj@hotspamnotmail.com> P Subject: Re: Will Rich Marcello either come clean or do some proper marketingand1 Message-ID: <dl6g0v$p6c$1@news-02.connect.com.au>   	 Hi Kerry,   L It doesn't appear that you're listening to me anymore but for the benefit ofF myself (if no one else) I can't let the following complete bollocks go unchallenged: -   ? >  In OpenVMS terms, that would include using technologies such B > as RTR which, similar to HW fault tolerance, ensures a committed > transaction is never lost.  - Did you mis-type RTR and were meaning DECdtm?   L If not were you talking about RTR's store-and-forward or guaranteed-deliveryK functionality when you were talking about a "transaction is never lost"? If 7 it's the latter then could you please answer me this: -   L What happens if the first part of the RTR transaction is give Fred 1 millionI dollars and the second part is take 1 million dollars from Jack. Now when J the transaction was started Jack had a million bucks but since the networkI link to the remote database holding Jack's account-details has been down, J he's transferred that 1M squid to his bookie. So RTR can stamp up and downJ as much as it wants with its "I want a million dollars!" (don't we all :-)I but it's just not gonna get it. Excellent stuff about the transaction not E being lost though! The ability to replay stale info is quite a trick.   I Look into my eyes, ACID is just-like-sooo-yesterday. It's a hangover from @ mainframe legacy days that simply cannot model today's web basedK requirements. . . You are getting sleepy . . . OMX Click uses it so it must C work for everyone else. . .You eyelids are getting heavy. . .if web J transactions didn't work then how does Amazon function? Hey? Hey? . . .You8 are in your happy place. . .no one can hurt you anymore.  	 WAKEUP!!!    Regards Richard Maher     2 "Main, Kerry" <Kerry.Main@hp.com> wrote in messageL news:FD827B33AB0D9C4E92EACEEFEE2BA2FB70CC3D@tayexc19.americas.cpqcorp.net...   > -----Original Message-----6 > From: JF Mezei [mailto:jfmezei.spamnot@teksavvy.com]! > Sent: November 12, 2005 6:31 PM  > To: Info-VAX@Mvb.Saic.Com > > Subject: Re: Will Rich Marcello either come clean or do some > proper marketingand  >  > Larry Kilgallen wrote:< > > Like the requirement to run on specialized hardware much > more expensive" > > than that used for VMS today ? >  > ; > VMS can run on workstations, it can run on wildfire class  > machines with 0 > the galaxy interface between instances of VMS. > E > VMS had fault tolerance, but that didn't prevent it from running on 9 > platforms that didn't have the fault tolerant hardware.  > < > Supporting NSK features would allow the NSK apps to run on > VMS, but they ; > would still want to make falt tolerant hardware for those  > applications.  >   E I guess that would depend on whether you thought the future for fault A tolerant computing was going to be based on hardware or software.   E The NSK folks have already stated they will be using standard Itanium C systems in a triplicated configuration that is essentially software % based to achieve its fault tolerance.   E Imho, there are just to many ways to achieve fault tolerant solutions H with software strategies than to design very high end expensive hardwareG solutions. In OpenVMS terms, that would include using technologies such @ as RTR which, similar to HW fault tolerance, ensures a committed transaction is never lost.   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)   4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sat, 12 Nov 2005 23:34:48 -0500 ( From: Bill Todd <billtodd@metrocast.net>P Subject: Re: Will Rich Marcello either come clean or do some proper marketingandG Message-ID: <o6ydnWVvtdf0X-venZ2dnUVZ_tadnZ2d@metrocastcablevision.com>    Main, Kerry wrote:   ...   G > The NSK folks have already stated they will be using standard Itanium E > systems in a triplicated configuration that is essentially software ' > based to achieve its fault tolerance.  > G > Imho, there are just to many ways to achieve fault tolerant solutions J > with software strategies than to design very high end expensive hardware > solutions.  F Oh, really?  Please list some that will catch 'quiet' hardware errors I (hardware-level operations which by all appearances execute successfully  : but in fact corrupt the system) of the form that hardware H comparator-based solutions (such as the traditional lock-step approach)  virtually guarantee to catch.   D The only one I know of involves software comparisons at every point 9 where the persistent state of the system changes *or any  H externally-visible result is returned* (persistent or not, because once G any information escapes from the system into the larger world, if it's  I incorrect a visible fault has occurred).  This is a) considerably slower  @ than a comparable hardware solution and b) not much of any less @ expensive (that is, if the hardware already supports a hardware G solution, as previous and previously-planned future NSK platforms did).   C So either Tandem is quietly bastardizing a previously bullet-proof  G product line at the behest of its bean-counter masters and hoping that  F no one will notice, or there's something very technically interesting D that the world should know about (I'm not going to bother searching I myself, because it's your job not only to know about it but to make such   useful information public).    - bill   ------------------------------  % Date: Sun, 13 Nov 2005 00:55:06 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> P Subject: Re: Will Rich Marcello either come clean or do some proper marketingand, Message-ID: <4376D51D.2CCE330E@teksavvy.com>   "Main, Kerry" wrote:G > The NSK folks have already stated they will be using standard Itanium E > systems in a triplicated configuration that is essentially software ' > based to achieve its fault tolerance.     # When was this announced ? In 2004 ?   > At the time of the merger completion when the state of VMS wasH announced'  I was explicitely told that the NSK group woudl have its ownH dedicated specialised IA64 hardware that was not related to the hardware; that VMS would be used (the later being shared with HP-UX).   E Or do you just mean that they will be using standard chipsets and CPU B boards, used in system chassis with all the specialised redundancy
 features ?   ------------------------------  % Date: Sat, 12 Nov 2005 18:06:04 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> X Subject: Re: Will Rich Marcello either come clean or do some proper marketingand PR PRPR, Message-ID: <43767558.DDEB749C@teksavvy.com>   Bill Gunshannon wrote:F > More of the same.  The processor battle is not over VMS or NSK.  YouF > really need to get a grip on reality.  Outside of comp.os.vms VMS isG > just not that important.  I imagine NSK is the same in it's circle of 
 > supporters.   G NSK is different from VMS in that regards. It truly is mission critical H in high visibility applications such as running the NYSE and NASDAQ.  ItG is a low volume, low profit but high visiblity thing. Killing NSK would G immediatly reverberate throughout the investment community and drive HP 
 shares down.    C The smart thing to do would be to get VMS to support all of the NSK G functionality and port some of the major software and applications from % NSK to VMS. Then you could drop NSK.    F In the Compaq days, VMS was certaintly invisible, but its revenus were7 necessary to help Compaq subsidize its wintel products.   H VMS remains invisible for HP, but I am not sure if the revenus are stillA worthy of special attention. With the murder of Alpha, I strongly D suspect that there was a huge dip in revenus. While it may be growinH now, it might still be too small a number to warrant a lot of attention.  H So, yes, perhaps you are right and HP would have no qualms about killingH it.  If VMS management aren't courageous enough to tell HP that the portC of VMS to the 8086 would greatly increase revenus, then perhaps the G customers could do so, very publically. This way, HP would be forced to & respond to the media and shareholders.   ------------------------------  % Date: Sat, 12 Nov 2005 19:45:25 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> Y Subject: Re: Will Rich Marcello either come clean or do some proper marketingandPR market , Message-ID: <43768C9A.9DC30955@teksavvy.com>   Bill Todd wrote:3 > > Intel has nothing serious running on its 8086s.  > J > You just mentioned one yourself:  Solaris.  Strike three, and you're out
 > of here.  @ The 32 bit Solaris may run on intel based systems, but Sun has aB relationship with AMD, builds its 64 bit systems with AMD, and the marketing is done with AMD.    AMD gets the bragging rights.   F Similarly, when Apple decided to go to the 8086, it struck a deal with/ Intel. Intel gets bragging rights, AMD doesn't.   A With commodity industry standard stuff, it is all about image and 
 marketing.   ------------------------------   End of INFO-VAX 2005.633 ************************                                                                                                                    4X-;y<Y?;6o|u]'wBn.'b'_<ـ02ASPN"F3Yoo*\a&MBX$jU,^&ҒEȻ̝0,o0"Qw^֛ޠ8kڎd3]2[pc>u~\ }7A:=ާ=jq887(mb×Bu eME}QxJD*4V Qe[/G/pX{n%42V0瓵(;PG:ډe0&=>4}%do
j#fr8Caat 7d(
 hֈg+'qU\u-D62'b7ˣǄdt!Cw;mM:qGHI౛.3|VH /u%k-Tq3N!`[.br,+5#L){!vBuMU(*h0?}Q(as
1ApD_:S*F
<)Z4Xe_F$b-8ꉄ%8~/-J9E
{8a@S~(o\3	䰼/.W]:c
q6!Jqtxvz%bI|s!b~/g~љHUL]L:WWcpOvNN w'0w}yk|$lX^WP;XA$xQWW9MC/zTZoܖ;էQŴJPPȢo9lբ(V{qD50ZoQ9mKaW[
iΫsVxʴ9<:n98z.;{J|/`:5C<>xv~%L(6a"1brp(Mnqƍ@yuk0lb+	|ԔqEG=`^9S$נN.8Akpn6d}!ϑт`w
$Ti"=6GUR0GqH&7:oG3.ohq&'"F唺/oބ1ޅ*olĕs@0:p"/W/؄j_n)tWa U-Dcnv% r@5Y[bl>*>P ~{f!3g-:im266N^[]]
O%\t`T5W^Bt.I 1mz'X_Y$UCuh]gnqށ"xm"Sl2>)}6dc`\̻ꌥ/Ց*[s-9{";j0TBlGȄ(.~<54tc?6qK=