home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!usc!howland.reston.ans.net!paladin.american.edu!auvm!ROHVM1.BITNET!MBDMD
- Message-ID: <L-VMCTR%93011208154739@VM1.CC.UAKRON.EDU>
- Newsgroups: bit.listserv.l-vmctr
- Date: Tue, 12 Jan 1993 08:15:47 -0500
- Sender: VMCENTER Components Discussion List <L-VMCTR@AKRONVM.BITNET>
- From: "Martin J. Doyle (215) 752-2296" <MBDMD@ROHVM1.BITNET>
- Subject: VMarchive dasd crash
- Lines: 82
-
- On Sat, 9 Jan 1993 03:57:06 GMT Stan Horwitz said:
- >...stuff deleted... exactly how much
- >trouble will we be in if this catalog file become corrupt or destroyed?
-
- Stan,
-
- If this data is so critical I think a DASD failure is the least of your
- problems. Have you taken a close look at VM/Archive's ability to
- create "twins". Most people skim over the sections on twins. They
- just create them and assume that they can be used for disaster recovery.
- Unfortunately, while a tape set is open, BOTH THE PRIMARY AND THE TWIN
- MUST be ON SITE! If you had a disaster that destroyed you tape library
- all of the data on the open tape sets would be lost forever since you
- had no way of creating an off site copy of the data.
-
- When people think about disaster recovery they concentrate on DASD and
- as they create tapes they create an on site and an off site copy. They
- forget the VM Archive tape pools are really just a logical extension
- of their DASD and must be recovered in a disaster. Your only options
- are to close out tape sets frequently (like every day), mark the twin
- as missing, and send it off site. You then have to go through merge
- purge on a regular basis and take the twin of the merge purge off site
- before you bring the other twins back. AN EXTREMELY MESSY PROCESS
- WHICH IS PRONE TO ERROR. The other option which we have chosen to
- implement is a tape duplication process. WARNING! The whole process
- is more complicated than just a tape copy. You need to generate a
- tape with the same VOLSER (this causes headaches when the tapes are
- returned to the scratch pool and have the wrong VOLSER). You also
- need to keep the HDRn files in tack. This can be tricky; we're willing
- to share our code to do this with anyone that's interested.
-
- Flame on...
-
- I can not believe that people are putting up with this problem,
- unfortunately, many people don't even know a problem exists. This
- problem is extremely serious. Lack of a REASONABLE method of disaster
- recovery jeopardizes the integrity of the data. If you had a fire that
- destroyed you tape vault today how much VM/Archive data would you
- loose? Most people that realize this problem exists do not put
- "production" data in VM/Archive because they can't afford to loose
- the data. That's really the key: DON'T STORE ANYTHING IN VM/ARCHIVE
- THAT YOU CAN'T AFFORD TO LOOSE.
-
- I can't believe that we're all paying these outrageous upgrade fees to
- move to VM/Archive/E. What are we getting for it? I understand that
- they are adding some interesting new function to the next release
- but what good is new function when there is a serious problem that
- jeopardizes the integrity of the data? Systems Center should always
- address all data integrity problems before they even think about adding
- new function.
-
- Systems Center knows about this problem. I've discussed it with them a
- number of times. It's a tough problem. I get the impression that they'd
- rather spend time adding function that sells more copies of the
- product instead of spending time to address integrity problems that
- many people are not aware of. This is not the way I'd treat my
- customers!
-
- If anyone is interested, I've been discussing this problem with the
- people at Syncsort. They sell a product that competes with VM/Archive.
- They use the same basic architecture as VM/Archive which means they
- have the exact same problem. The big difference is that SYNCSORT SEES
- THIS PROBLEM AS AN OPPORTUNITY TO MAKE THEIR PRODUCT BETTER THAN
- VM/ARCHIVE AND AN OPPORTUNITY TO TAKE MARKET SHARE. I doubt that
- Systems Center is going to do much about this problem until they
- realize that people are mad and are willing to talk to another vendor.
- If you are interested in discussing this problem with SYNCSORT just
- send me a note and I'll be glad to give you the names and numbers of
- the people at SYNCSORT who are addressing this issue.
-
- Flame off...
-
-
- Thanks,
-
- Martin J. Doyle
- VM Systems Programming Contractor
- Rohm and Haas Company
- Philadelphia, Pennsylvania
- MBDMD@ROHVM1
- (215) 592-6865 (Day)
- (215) 752-2296 (Evening)
-