home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!charon.amdahl.com!amdahl!JUTS!rbw00
- From: rbw00@ccc.amdahl.com ( 324 Richard Wilmot)
- Newsgroups: comp.databases
- Subject: Re: distributed transactions
- Message-ID: <01ZY02L01fUR01@JUTS.ccc.amdahl.com>
- Date: 23 Aug 92 04:05:58 GMT
- References: <1736k9INNjhh@agate.berkeley.edu> <BtCrA6.Iuy@cup.hp.com> <173oogINNmrp@agate.berkeley.edu> <BtDA1n.JrB@world.std.com>
- Reply-To: rbw00@JUTS.ccc.amdahl.com ( 324 Richard Wilmot)
- Organization: Amdahl Corporation, Sunnyvale CA
- Lines: 37
-
- In article <BtDA1n.JrB@world.std.com> edwards@world.std.com (Jonathan Edwards) writes:
- >My original post was a query to find out if any commercial databases
- >were capable of robust hot standby. So far, NOT. So far, there doesn't
- >even seem much acknowledgement that this is a correct approach
- >to the problem.
-
- Sorry, vacation prevented me seeing the original post. Remote
- distributed databases may not have a fail-over capability but
- IBM has the Extended Recovery Facility which allows their IMS/VS
- database system to take over for another failing IMS database
- system. BUT the recovering system must be cabled to the same
- database disk drives and logging (journaling) disk drives. The journal
- files may be duplicated (but still need to be accessible to both
- computer systems). The cabling can be out to 9 km (?? at home
- so don't have reference material).
-
- IMS is not a relational database but I expect IBM to equip their
- DB2 relational product also with Extended Recovery. I would also
- expect the distance to be wider as IBM changes disk protocols.
-
- Neither IMS nor DB2 run under Unix (TM). They need IBM's MVS operating
- system which runs on IBM, Amdahl, or Hitachi machines.
-
- Another requirement of Extended Recovery is that there be a double
- cabled communication controller. Special changes were required to
- communication protocols (e.g. don't ask user to log on to the alternate
- system as it takes over) and I would expect startup of the failover
- process to be extremely interesting: Are you (primary system) really
- dead? If processing 500 transactions at the time of a failure then
- 500 transactions will need to be backed out and restarted and more
- than 500 user sessions will need to be established before any user
- sees any new stuff on their screens. This should provide interesting
- times indeed.
-
- --
- Dick Wilmot | I declaim that Amdahl might disclaim any of my claims.
- (408) 746-6108
-