1 INFO-VAX	Thu, 10 Aug 2006	Volume 2006 : Issue 443       Contents: Re: Agile Manifesto (maybe OT), CIFS evaluation kit - SYSUAF authentication?% Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP % Re: Clients using a GUI to access FTP   Re: DECnet-IV strangeness in NCP  Re: DECnet-IV strangeness in NCP  Re: DECnet-IV strangeness in NCP+ External Authentication for Advanced Server / Re: External Authentication for Advanced Server / Re: External Authentication for Advanced Server @ Re: OpenVMS in the Press - Examples of Mission Critical Clusters@ RE: OpenVMS in the Press - Examples of Mission Critical Clusters@ RE: OpenVMS in the Press - Examples of Mission Critical Clusters Re: Speaking of Clusters:  Re: Speaking of Clusters: & Re: VMS backup: competition from Apple  F ----------------------------------------------------------------------    Date: 10 Aug 2006 04:41:18 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen) ' Subject: Re: Agile Manifesto (maybe OT) 3 Message-ID: <8b+zjJVO2YeM@eisner.encompasserve.org>   z In article <8660a3a10608091351q398837ct4298c69489adb9a4@mail.gmail.com>, "William Webb" <william.w.webb@gmail.com> writes:  C > Our highest priority is to satisfy the customer through early and + > continuous delivery of valuable software.   B If the software is valuable, "continuous delivery" of new versions2 is not a customer priority.  It adds to confusion.  @ > Welcome changing requirements, even late in development. AgileD > processes harness change for the customer's competitive advantage.  , To me this seems contrary to robust testing.   ------------------------------  # Date: Thu, 10 Aug 2006 06:59:35 GMT L From: winston@SSRL.SLAC.STANFORD.EDU (Alan Winston - SSRL Central Computing)5 Subject: CIFS evaluation kit - SYSUAF authentication? 6 Message-ID: <00A59F98.547C03AD@SSRL.SLAC.STANFORD.EDU>  0 VMS 8.2-1 on RX2600; evaluation version of CIFS.  A Is this the best place to ask questions about CIFS/Samba on VMS?    I Is there a way to get CIFS (Samba) to use SYSUAF authentication?  I'm not H talking external authentication - just let the user provide VMS username& and password to connect to the server.  N If that's supported, how do I specify it in SMB.CONF?  What are the drawbacks?N (The drawbacks of using SMBPASSWD to set it in a local database include havingH *yet another* password file to get out of synch with the VMS and Windows passwords.)   J While I'm asking, anybody got a better-defined release date than "H2" for I the first production release?  Are there any indications of what will be  I different in the production release?  It looks like this is something I'm 7 going to need to be using fairly seriously fairly soon.   1 Would I be a lot better off using a Samba 2 port?    Thanks,    -- Alan    ------------------------------  % Date: Thu, 10 Aug 2006 02:04:51 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> . Subject: Re: Clients using a GUI to access FTP, Message-ID: <44DACC7E.58F2320A@teksavvy.com>  F Note: old netscape 4.7 on a mac works fine with the tcpip services FTP server 5.3 vax).  H I get file names without any dates (which indicates it issues the proper NLIST command).     H Also, and this may make a big difference, you need to define the logical* that tells FTP to ignore version numbers.   % $define/system tcpip$ftp_no_version 1     E Since there are plenty of FTP clients that work fine with the VMS FTP G servers, I would point to those customers that it is only the microsoft S software which is incompatible because it, as usual, does not follow the standards.    ------------------------------  % Date: Thu, 10 Aug 2006 02:23:05 -0400 ' From: Dave Froble <davef@tsoft-inc.com> . Subject: Re: Clients using a GUI to access FTP9 Message-ID: <df6dndDdC6y0UkfZnZ2dnUVZ_q-dnZ2d@libcom.com>    JF Mezei wrote: H > Note: old netscape 4.7 on a mac works fine with the tcpip services FTP > server 5.3 vax). > J > I get file names without any dates (which indicates it issues the proper > NLIST command).  >  > J > Also, and this may make a big difference, you need to define the logical, > that tells FTP to ignore version numbers.  > ' > $define/system tcpip$ftp_no_version 1  >  > G > Since there are plenty of FTP clients that work fine with the VMS FTP I > servers, I would point to those customers that it is only the microsoft U > software which is incompatible because it, as usual, does not follow the standards.   F It's my understanding that there are no standards for this.  MS isn't D wrong, just pig headed.  We can be sure there are people at MS that I understand their product doesn't work with VMS.  They either don't care,   or like things that way.  I But I agree.  Test 5 browsers, and if 4 work, and IE doesn't, claim that   it's MS that's not 'correct'.    --  4 David Froble                       Tel: 724-529-0450> Dave Froble Enterprises, Inc.      E-Mail: davef@tsoft-inc.com DFE Ultralights, Inc.  170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------   Date: 10 Aug 2006 11:11:10 GMT( From: bill@cs.uofs.edu (Bill Gunshannon). Subject: Re: Clients using a GUI to access FTP+ Message-ID: <4k0iieFa2pu7U1@individual.net>   9 In article <df6dndDdC6y0UkfZnZ2dnUVZ_q-dnZ2d@libcom.com>, * 	Dave Froble <davef@tsoft-inc.com> writes: > K > But I agree.  Test 5 browsers, and if 4 work, and IE doesn't, claim that   > it's MS that's not 'correct'.   F Which is fine until the customer tries IE with 5 different systems andC VMS is the only one that doesn't work.  What will the customer say? 5 And what does the old proverb say about The Customer?    bill      --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------    Date: 10 Aug 2006 07:32:44 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) . Subject: Re: Clients using a GUI to access FTP3 Message-ID: <sT32kHOpqSHV@eisner.encompasserve.org>   l In article <00cb01c6bbc4$0a1e7470$2802a8c0@CHARONVAX>, "Peter Weaver" <info-vax@weaverconsulting.ca> writes: > L > I don't know, all I know is that I had a customer complaining that my FTP M > server is broken because HP knows that MS ignored the RFCs when they wrote  F > IE but HP stands firm and tells us to complain to MS. If I remember M > correctly Process provides a logical that can be set to tell their TCPWare  I > FTP server to present listings in a Unix way just for this problem. HP  J > should be able to do the same thing instead of telling us that they are  > right and MS is wrong  >   F    Once upon a time VAX Fortran was the defacto industry standard.  HPF    and Sun didn't tell prospective customers they we're trying to lureI    away from DEC that all that extra stuff wasn't ANSI standard Fortran,  =    they changed their compilers to be VAX Fortran compatable.   G    UNIX ls output actually differs between SYSV and BSD, but has become :    the defacto industry standard for the FTP LIST command.  D    And once again UCX rears its head as the fellow behind the curve,/    renaming it TCP/IP Services didn't fix that.    ------------------------------    Date: 10 Aug 2006 07:35:56 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) . Subject: Re: Clients using a GUI to access FTP3 Message-ID: <DS$AGVdmNfFZ@eisner.encompasserve.org>   c In article <df6dndDdC6y0UkfZnZ2dnUVZ_q-dnZ2d@libcom.com>, Dave Froble <davef@tsoft-inc.com> writes:   H > It's my understanding that there are no standards for this.  MS isn't F > wrong, just pig headed.  We can be sure there are people at MS that K > understand their product doesn't work with VMS.  They either don't care,   > or like things that way.  E    Considering the terminal emulator that ships with Windows has been B    broken since at least W95 and MS doens't care, I can hardly see;    compatability with VMS in IE as getting thier attention.   @    Getting MS to follow documented standards went out the window    under MS-DOS.  %    Which almost looks like a bad pun.    ------------------------------   Date: 10 Aug 2006 12:56:19 GMT( From: bill@cs.uofs.edu (Bill Gunshannon). Subject: Re: Clients using a GUI to access FTP+ Message-ID: <4k0onjF7u30rU1@individual.net>   3 In article <sT32kHOpqSHV@eisner.encompasserve.org>, > 	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:n > In article <00cb01c6bbc4$0a1e7470$2802a8c0@CHARONVAX>, "Peter Weaver" <info-vax@weaverconsulting.ca> writes: >>  M >> I don't know, all I know is that I had a customer complaining that my FTP  N >> server is broken because HP knows that MS ignored the RFCs when they wrote G >> IE but HP stands firm and tells us to complain to MS. If I remember  N >> correctly Process provides a logical that can be set to tell their TCPWare J >> FTP server to present listings in a Unix way just for this problem. HP K >> should be able to do the same thing instead of telling us that they are   >> right and MS is wrong >>   > H >    Once upon a time VAX Fortran was the defacto industry standard.  HPH >    and Sun didn't tell prospective customers they we're trying to lureK >    away from DEC that all that extra stuff wasn't ANSI standard Fortran,  ? >    they changed their compilers to be VAX Fortran compatable.  > I >    UNIX ls output actually differs between SYSV and BSD, but has become < >    the defacto industry standard for the FTP LIST command.  E How do they differ?  "ls" returns a simple list of filenames in both. E "ls -l" returns; file modes, number of links, owner, group, filesize, H last modified time and filename.  While beyond that they offer differentB options these are the only two generic to a discussion of what FTPD returns.  And, while most (all?) unix systems use the ls command andF pipe the output thru the FTP server it seems it would have been ratherC trivial for VMS to read the directory and send the information in a D similar (and compatable) format.  Whyh they don't is not information I am privy to.   bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  # Date: Thu, 10 Aug 2006 13:49:49 GMT , From: Hoff Hoffman <hoff-remove-this@hp.com>. Subject: Re: Clients using a GUI to access FTP1 Message-ID: <1QGCg.1833$lL6.724@news.cpqcorp.net>    Bill Gunshannon wrote:  G > How do they differ?  "ls" returns a simple list of filenames in both. G > "ls -l" returns; file modes, number of links, owner, group, filesize, J > last modified time and filename.  While beyond that they offer differentD > options these are the only two generic to a discussion of what FTPF > returns.  And, while most (all?) unix systems use the ls command andH > pipe the output thru the FTP server it seems it would have been ratherE > trivial for VMS to read the directory and send the information in a F > similar (and compatable) format.  Whyh they don't is not information > I am privy to.  H    This then starts to look like the C headers, and all the conditional 1 logic that builds up in such as the targets move.   F    And as for mapping OpenVMS names into Unix names (or the other way G 'round), there are a gazillion obvious filename translations, and some  H really ugly ones.  What's ., after all?  The corner cases are enough to $ drive any sane engineer up the wall.  D    The only way I can see to reasonably engineer this within TCP/IP D Services likely involves a sniff of the client, and add a flag that F resets the output to look like Unix.  (This so we don't bust existing G clients that expect existing formats.)  But I've seen cases where this  I won't work, because the client sniffs the server, and reacts differently  G depending on what it finds.  Which leads to ever-lengthening stacks of  H hackery, and more control knobs, and more misleading responses when you F ask for identification, and pretty soon you end up with.... SCSI!  :-)  I    Alternatively, to update the FTP RFC to add a new FTP sequence with a  F defined format for the filenames and directory names, and to wait for @ everybody to agree to it and to decide to implement and ship it.  H    Yes, I'd like to see a way for Microsoft and TCP/IP Services to play G together better here, but then I also gave up on Internet Explorer and  G have been using Mozilla Firefox for eons now.  I do need to test-drive  8 the newest IE7 stuff, to see how much that has improved.   ------------------------------    Date: 10 Aug 2006 07:02:16 -0700 From: etmsreec@yahoo.co.uk. Subject: Re: Clients using a GUI to access FTPC Message-ID: <1155218536.468156.311420@m79g2000cwm.googlegroups.com>   C Guess that'd be the same Microsoft that HP buy so much from and (in B their earlier guises as DEC and Compaq) allowed to take so much of their traditional business.   D "Industry-standard" works, except where you want to break it or just not comply with it.    Steve    Peter Weaver wrote: N > I know this topic has come up in the past and HP has said that IE is broken,M > but that answer is like the guy being pulled out of his crumpled car by the J > ambulance attendants saying "Sure I saw the transport truck coming at me6 > with its brakes on fire; But I had the green light." > I > So has there been any advancement on having the "HP TCP/IP Services for N > OpenVMS Alpha Version V5.5" product produce a listing that IE would be happyJ > with? I distribute files to my clients from my VMS box using FTP and one9 > client complained that he could not use IE (clicking on K > ftp://username:password@ftp.weaverconsulting.ca) to access my files so my J > site MUST be broken since IE works with every other FTP site he has ever
 > been on. > M > So fine, HP is right, Microsoft is wrong, but I'm stuck walking my customer N > through using the DOS FTP client because HP would rather be right than work.L > Very embarrassing for someone trying to show the world that VMS is a great > product to use.  >  > Peter Weaver > www.weaverconsulting.ca  > CHARON-VAX  CHARON-AXP   ------------------------------   Date: 10 Aug 2006 14:59:25 GMT( From: bill@cs.uofs.edu (Bill Gunshannon). Subject: Re: Clients using a GUI to access FTP+ Message-ID: <4k0vudF9v06bU1@individual.net>   1 In article <1QGCg.1833$lL6.724@news.cpqcorp.net>, / 	Hoff Hoffman <hoff-remove-this@hp.com> writes:  > Bill Gunshannon wrote: > H >> How do they differ?  "ls" returns a simple list of filenames in both.H >> "ls -l" returns; file modes, number of links, owner, group, filesize,K >> last modified time and filename.  While beyond that they offer different E >> options these are the only two generic to a discussion of what FTP G >> returns.  And, while most (all?) unix systems use the ls command and I >> pipe the output thru the FTP server it seems it would have been rather F >> trivial for VMS to read the directory and send the information in aG >> similar (and compatable) format.  Whyh they don't is not information  >> I am privy to.  > J >    This then starts to look like the C headers, and all the conditional 3 > logic that builds up in such as the targets move.   H I don't see this.  I think you have misunderstood what I was suggesting.   > H >    And as for mapping OpenVMS names into Unix names (or the other way I > 'round), there are a gazillion obvious filename translations, and some  J > really ugly ones.  What's ., after all?  The corner cases are enough to & > drive any sane engineer up the wall.  H I wouldn't suggest mapping VMS names to Unix names.  Why would you?  AllG your VMS names are legitimate Unix names any way (even with the ";13"). F If the VMS directory doesn't contain a entry called "." you just don't4 include one.  "ls" doesn't even though there is one.   > F >    The only way I can see to reasonably engineer this within TCP/IP F > Services likely involves a sniff of the client, and add a flag that H > resets the output to look like Unix.  (This so we don't bust existing * > clients that expect existing formats.)    D Now I know I am confused.  I thought the problem was GUI Ftp ClientsC interpreting the information from a "DIR" command to an FTP Server. E All that means is making the VMS FTP Server send back lines that look  like: 7 -rw-r--r--  2 bill  120   3235 Sep 20  2001 test01a.cob 9 -rw-r--r--  1 bill  120   3097 Sep 20  2001 test01a.cob;1 9 -rw-r--r--  1 bill  120   3227 Sep 20  2001 test01a.cob;2 9 -rw-r--r--  2 bill  120   3235 Sep 20  2001 test01a.cob;3 7 -rwxr-x---  1 bill  120   4608 Sep 20  2001 test01a.exe   E Note that some of these files actually have version numbers.  Care to  guess where that came from?   H >                                        But I've seen cases where this K > won't work, because the client sniffs the server, and reacts differently  I > depending on what it finds.  Which leads to ever-lengthening stacks of  J > hackery, and more control knobs, and more misleading responses when you H > ask for identification, and pretty soon you end up with.... SCSI!  :-)  F Some may do that, but my experience has been that most GUI Ftp ClientsC just work with the human readable output and as long as you provide E something that looks like iwhat comes from most servers it will work.    > K >    Alternatively, to update the FTP RFC to add a new FTP sequence with a  H > defined format for the filenames and directory names, and to wait for B > everybody to agree to it and to decide to implement and ship it.  D Again, I don't think that is the problem.  There is nothing in a VMSH filename that is not valid in a Unix filename so no re-mapping of actual names would be needed.   > J >    Yes, I'd like to see a way for Microsoft and TCP/IP Services to play I > together better here, but then I also gave up on Internet Explorer and  I > have been using Mozilla Firefox for eons now.  I do need to test-drive  : > the newest IE7 stuff, to see how much that has improved.  F But the fact is the majority of the world is using MS and if you don'tE interact with it then you become the odd man out.  I am not sure that 1 is a postion that VMS can effectively work from.     bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------   Date: 9 Aug 2006 23:24:45 -0700 1 From: "Bart.Zorn@gmail.com" <Bart.Zorn@gmail.com> ) Subject: Re: DECnet-IV strangeness in NCP B Message-ID: <1155191085.033746.298920@75g2000cwc.googlegroups.com>   Hoff Hoffman wrote:  > VAXelate wrote: D > > Alpha ES40 running 7.2-1 and phase IV DECnet.  It is functioningK > > properly, but NCP just recently started showing VERY strange behaviour. J > >  If I do an NCP SHOW KNOWN NODES or NCP SHOW NODE 2.*, it doesn't showH > > any Area 2 nodes, but if I specifically do an NCP SHOW NODE 2.50 (orC > > any other normally defined Area 2 node), it displays correctly:  > H >    DECnet Phase IV isn't, AFAIK, set up as a router on DECnet Phase IVE > on OpenVMS Alpha (host-based routing was added for DECnet-Plus), so E > you're probably operating as an end-node or as an extended-function H > (cluster alias) host.  Or are you executing this NCP command on a node3 > configured as a DECnet level 1 or level 2 router?  > J >    Are you using a local database or are you using DNS (DECdns) for name
 > service?  ( DECdns with DECnet Phase IV? Not likely!   >  >    SHOW EXEC CHAR, please? > # >    Does SET KNOWN NODES ALL help?  > ' >    Here's the current SPD for DECnet:  > 4 > http://h18000.www1.hp.com/info/SP4848/SP4848PF.PDF  	 Bart Zorn    ------------------------------    Date: 10 Aug 2006 04:44:06 -0700% From: "VAXelate" <moore.mc@gmail.com> ) Subject: Re: DECnet-IV strangeness in NCP A Message-ID: <1155210246.793612.35050@i3g2000cwc.googlegroups.com>    Bart.Zorn@gmail.com wrote:   > * > DECdns with DECnet Phase IV? Not likely! >  > >  > >    SHOW EXEC CHAR, please? > > % > >    Does SET KNOWN NODES ALL help?  > > ) > >    Here's the current SPD for DECnet:  > > 6 > > http://h18000.www1.hp.com/info/SP4848/SP4848PF.PDF >  > Bart Zorn   F Good point - that's only OSI isn't it? (only a few Phase V boxes here,: and no DECdns among them, so I don't have that experience)  G I executed an NCP SET KNOWN NODES ALL and it didn't seem to help at the G time, however some hours later last night, the problem had mysteriously G "corrected itself".  Don't know if the NCP command helped or possibly a C routing table update cleared whatever was getting in the way of the  display?  D If it was on Windows, I would just presume "sometimes that happens",C but something about OpenVMS (its bulletproof nature? its (normally) 0 repeatable behaviour?) leads me to want to know:  F What caused the "Disappearing inconsistent Area 2 node display"?   and What actually fixed it?    ------------------------------  # Date: Thu, 10 Aug 2006 12:59:05 GMT , From: Hoff Hoffman <hoff-remove-this@hp.com>) Subject: Re: DECnet-IV strangeness in NCP 0 Message-ID: <t4GCg.1830$4G6.80@news.cpqcorp.net>   Bart.Zorn@gmail.com wrote: > Hoff Hoffman wrote:   * > DECdns with DECnet Phase IV? Not likely!  I    Ran it all the time, and for many years.  Worked fine, so long as the  E network and the DNS servers were themselves holding stable.  NCP SET  H EXEC DNS, et al.  The DECdns capability had been latent for a while and I appeared (quietly) back around OpenVMS VAX V5.5-2...  (I see it is still  G commented out in the OpenVMS Alpha help library text for NCP, however.)       --   I    This is not the DNS/BIND name service seen commonly with IP networks,  F this is a different name service known variously as DNS and as DECdns.   ------------------------------    Date: 10 Aug 2006 08:45:39 -0700' From: "syslost" <wm.reynolds@gmail.com> 4 Subject: External Authentication for Advanced ServerB Message-ID: <1155224739.436686.248220@i3g2000cwc.googlegroups.com>  4 Two ds25 Alpha OpenVMS V7.3-1 NI cluster, hsg80 SAN.  E We are using Advanced Server (Pathworks v7.3a), and I have been asked D if we can use external authentication to a Windows server.  I know IG know, rtfm.  For a short answer, is it possible?  And any gotyas others  have run into.   thanks in advance.   ------------------------------    Date: 10 Aug 2006 09:22:56 -0700  From: "Ian Miller" <ijm@uk2.net>8 Subject: Re: External Authentication for Advanced ServerC Message-ID: <1155226976.666679.208720@m79g2000cwm.googlegroups.com>    possibly  M http://h71000.www7.hp.com/doc/73final/6556/6556pro_007.html#users2_enabextaut   : Are you using ActiveDirectory and what version of windoze?   ------------------------------    Date: 10 Aug 2006 10:01:12 -0700' From: "syslost" <wm.reynolds@gmail.com> 8 Subject: Re: External Authentication for Advanced ServerC Message-ID: <1155229272.761355.247610@q16g2000cwq.googlegroups.com>   8 Thanks, for the link!! I looking at it right now.  No to@ ActiveDirectory, and 2003 on the Servers, and XP on the desktop.     Ian Miller wrote: 
 > possibly > O > http://h71000.www7.hp.com/doc/73final/6556/6556pro_007.html#users2_enabextaut  > < > Are you using ActiveDirectory and what version of windoze?   ------------------------------  # Date: Thu, 10 Aug 2006 11:46:04 GMT " From:   VAXman-  @SendSpamHere.ORGI Subject: Re: OpenVMS in the Press - Examples of Mission Critical Clusters 0 Message-ID: <00A59FD9.7F02C218@SendSpamHere.ORG>  ~ In article <FA60F2C4B72A584DBFC6091F6A2B868401873FE0@tayexc19.americas.cpqcorp.net>, "Main, Kerry" <Kerry.Main@hp.com> writes: >  >  >All,  > C >Readers of this list might be interested in the following links as 3 >examples of mission critical, multi-site clusters:  > D >1. CareTech Solutions Completes Critical Infrastructure Upgrades inC >Record Time for Detroit Medical Center's Electronic Medical Record  >Initiative H >http://www.caretechsolutions.com/default.aspx?id=3D19&sid=3D1&pr=3D58 = >(Aug 8, >2006)  G Will you please tinyURL these links so that your M$ software can't hork  them up when posted?   --  K VAXman- A Bored Certified VMS Kernel Mode Hacker   VAXman(at)TMESIS(dot)COM              5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------  % Date: Thu, 10 Aug 2006 09:43:10 -0400 ' From: "Main, Kerry" <Kerry.Main@hp.com> I Subject: RE: OpenVMS in the Press - Examples of Mission Critical Clusters T Message-ID: <FA60F2C4B72A584DBFC6091F6A2B868401874175@tayexc19.americas.cpqcorp.net>   > -----Original Message-----E > From: VAXman-@SendSpamHere.ORG [mailto:VAXman-@SendSpamHere.ORG]=20  > Sent: August 10, 2006 7:46 AM  > To: Info-VAX@Mvb.Saic.Com < > Subject: Re: OpenVMS in the Press - Examples of Mission=20 > Critical Clusters  >=20 > In article=20 @ > <FA60F2C4B72A584DBFC6091F6A2B868401873FE0@tayexc19.americas.cp7 > qcorp.net>, "Main, Kerry" <Kerry.Main@hp.com> writes:  > >  > >  > >All,  > > E > >Readers of this list might be interested in the following links as 5 > >examples of mission critical, multi-site clusters:  > > F > >1. CareTech Solutions Completes Critical Infrastructure Upgrades inE > >Record Time for Detroit Medical Center's Electronic Medical Record 
 > >Initiative D > >http://www.caretechsolutions.com/default.aspx?id=3D3D19&sid=3D3D1 > &pr=3D3D58 =3D
 > >(Aug 8, > >2006) >=20A > Will you please tinyURL these links so that your M$ software=20  > can't hork > them up when posted? >=20 > --=20 7 > VAXman- A Bored Certified VMS Kernel Mode Hacker  =20  > VAXman(at)TMESIS(dot)COM >           =20 9 >   "Well my son, life is like a beanstalk, isn't it?"=20  >=20   My M$ software?   - Hey, perhaps I should ask about royalties eh?    :-)    CareTech Solutions:  http://tinyurl.com/orzox   Eurex Exchange Upgrades: http://tinyurl.com/rryeo   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: Thu, 10 Aug 2006 16:32:33 GMT " From:   VAXman-  @SendSpamHere.ORGI Subject: RE: OpenVMS in the Press - Examples of Mission Critical Clusters 0 Message-ID: <00A5A001.84835533@SendSpamHere.ORG>  ~ In article <FA60F2C4B72A584DBFC6091F6A2B868401874175@tayexc19.americas.cpqcorp.net>, "Main, Kerry" <Kerry.Main@hp.com> writes: >  >My M$ software?  F Well, whatever you're using to post is putting your posting text into  quoted-printable.      --  K VAXman- A Bored Certified VMS Kernel Mode Hacker   VAXman(at)TMESIS(dot)COM              5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------  % Date: Thu, 10 Aug 2006 01:56:27 -0400 ( From: Bill Todd <billtodd@metrocast.net>" Subject: Re: Speaking of Clusters:G Message-ID: <ze-dnSpvnJsRV0fZnZ2dnUVZ_s6dnZ2d@metrocastcablevision.com>    Keith Parris wrote: ! > david20@alpha2.mdx.ac.uk wrote: H >> Never having used AIX I didn't recognise HACMP and made the erroniousG >> assumption from context that it was another Linux failover solution. J >> I'm quite happy to accept your word that it is a good cluster solution. > K > HACMP is a run-of-the-mill UNIX High-Availability (HA) cluster solution,  F > roughly comparable to what TruCluster could do back in the ASE days 
 > (pre-V5.0).   G Not when you include available extensions like GPFS and the HACMP DLM,  	 it isn't.   =   Focus is all on failover. You see the typical references to ? > heartbeat, failover scripts, and active and "takeover" nodes.  > H > It's interesting that IBM must define the term "high availability" to A > include some implied downtime, as their solution can't provide  B > continuous application availability like an OpenVMS cluster can.J > (From "High Availability Cluster Multi-Processing for AIX: Concepts and H > Facilities Guide" http://publib.boulder.ibm.com/epubs/pdf/c2348646.pdfJ > "The difference between fault tolerance and high availability, then, is J > this: a fault tolerant environment has no service interruption, while a H > highly available environment has a minimal service interruption. Many B > sites are willing to absorb a small amount of downtime with highG > availability rather than pay the much higher cost of providing fault  I > tolerance. ... High availability systems are an excellent solution for  H > applications that can withstand a short interruption should a failure K > occur, but which must be restored quickly." By IBM's definition, OpenVMS  G > Clusters achieve Fault Tolerance, not just High Availability (and in  C > practice they actually do so at no additional cost beyond what's  " > required for High Availability.)  I I suspect you just don't understand what IBM is saying:  their reference  H to 'short interruption' appears to be to fail-over latency (measured in D seconds, just as VMS cluster state transitions are), rather than to H 'down time' per se (i.e., actual periods when a client sees the service E as unavailable - not there at all - rather than merely brief service   delays as a transition occurs).   H IBM's definition of 'fault tolerance' is also something which you don't H appear to understand:  it refers to *zero*-latency interruptions when a H fault occurs - because a second system executing the *same workload* in G parallel just picks up the workload instantly as the failing system is  G fenced out.  This is not, of course, what a VMS cluster provides - you  F need to look to systems like Tandem's or Parallel Sysplex to find it, D though IIRC Sun has products in that area and some even exist for - 8 gasp! - Windows (Stratus, still, and perhaps Marathon?).   - bill   ------------------------------    Date: 10 Aug 2006 09:01:47 -0700- From: "Andrew" <andrew_harrison@symantec.com> " Subject: Re: Speaking of Clusters:C Message-ID: <1155225706.935470.296130@b28g2000cwb.googlegroups.com>    Bill Todd wrote: > Keith Parris wrote: # > > david20@alpha2.mdx.ac.uk wrote: J > >> Never having used AIX I didn't recognise HACMP and made the erroniousI > >> assumption from context that it was another Linux failover solution. L > >> I'm quite happy to accept your word that it is a good cluster solution. > > L > > HACMP is a run-of-the-mill UNIX High-Availability (HA) cluster solution,G > > roughly comparable to what TruCluster could do back in the ASE days  > > (pre-V5.0).  > H > Not when you include available extensions like GPFS and the HACMP DLM, > it isn't.  > ? >   Focus is all on failover. You see the typical references to A > > heartbeat, failover scripts, and active and "takeover" nodes.  > > I > > It's interesting that IBM must define the term "high availability" to B > > include some implied downtime, as their solution can't provideD > > continuous application availability like an OpenVMS cluster can.K > > (From "High Availability Cluster Multi-Processing for AIX: Concepts and J > > Facilities Guide" http://publib.boulder.ibm.com/epubs/pdf/c2348646.pdfK > > "The difference between fault tolerance and high availability, then, is K > > this: a fault tolerant environment has no service interruption, while a I > > highly available environment has a minimal service interruption. Many D > > sites are willing to absorb a small amount of downtime with highH > > availability rather than pay the much higher cost of providing faultJ > > tolerance. ... High availability systems are an excellent solution forI > > applications that can withstand a short interruption should a failure L > > occur, but which must be restored quickly." By IBM's definition, OpenVMSH > > Clusters achieve Fault Tolerance, not just High Availability (and inD > > practice they actually do so at no additional cost beyond what's$ > > required for High Availability.) > J > I suspect you just don't understand what IBM is saying:  their referenceI > to 'short interruption' appears to be to fail-over latency (measured in E > seconds, just as VMS cluster state transitions are), rather than to I > 'down time' per se (i.e., actual periods when a client sees the service F > as unavailable - not there at all - rather than merely brief service! > delays as a transition occurs).  > I > IBM's definition of 'fault tolerance' is also something which you don't I > appear to understand:  it refers to *zero*-latency interruptions when a I > fault occurs - because a second system executing the *same workload* in H > parallel just picks up the workload instantly as the failing system isH > fenced out.  This is not, of course, what a VMS cluster provides - youG > need to look to systems like Tandem's or Parallel Sysplex to find it, E > though IIRC Sun has products in that area and some even exist for - : > gasp! - Windows (Stratus, still, and perhaps Marathon?). > F I would agree, neither HACMP or OpenVMS clusters meet IBM's definitionF of fault tolerance because both of them have failover or cluster stateC transition times that are measured in seconds rather than subsecond C switching to a second system/component running the same code at the F same point in the code as the failed component/system which you see in? an FT system. In addition because the FT systems are running in G hardware or software lock-step the switching time is very well defined, E this is not the case with failover or CST. Sure CST is within a range F of values but this is not good enough for some classes of application.  D In telco environments Tandem and Stratus were commonplace componentsE providing an interface between the network switches using SS7 and the F rest of the computing infrastructure. Without a FT device connected to@ the switch SS7 traffic which might include important CDR billing; information could be dropped resulting loss or revenue etc.   F Conventional cluster based high availability solutions such as OpenVMSG Clustering or HACMP could not meet this requirment because the failover C time/CST time was too long and would result in dropped SS7 traffic.   B In practice the demand for FT systems has declined in part becauseE middleware has become more sophisticated, replicating enough state to B ensure that if a failure happens it happens with minimal impact to clients.  F Bill mentioned Sun but most of the J2EE environments including the SunF one which uses the Clustra product have the ability to replicate stateE information across the nodes in the application server infrastructure D so that when a app server fails the clients are not impacted. CoupleF this with Oracle RAC configured for HA at the back end and you have anG software and hardware infrastructure that has some of the attributes of  a FT system.     Regards  Andrew Harrison    ------------------------------    Date: 10 Aug 2006 07:23:18 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) / Subject: Re: VMS backup: competition from Apple 3 Message-ID: <O6XCJktCe0iD@eisner.encompasserve.org>   H In article <ebdg10016pq@enews3.newsguy.com>, healyzh@aracnet.com writes: > L > There was no transition for the OS itself, Mac OS X started out running onN > x86.  The original developers releases ran on x86.  The port was to PPC, notM > x86.  The x86 port was done well prior to the PPC port, and was from 68k to I > x86.  But then since you can't grasp the fact that modern CPU's are not K > 8086's, I shouldn't expect anything you post to make sense, or for you to K > even remotely have your facts straight.  Do you do anything besides troll 
 > anymore?  G    Parts of the OS were running on x86, but not all of it, most notably A    the GUI.  Parts of a Mac OS without a the GUI is not a Mac OS.    ------------------------------   End of INFO-VAX 2006.443 ************************