home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: bit.listserv.db2-l
- Path: sparky!uunet!psinntp!cmcl2!panix!awerman
- From: awerman@panix.com (Aaron Werman)
- Subject: Re: DB2 Back up procedure
- Message-ID: <C1D1rw.J2s@panix.com>
- Organization: PANIX Public Access Unix, NYC
- References: <930124.091333.SAT.OPRIMS@SAUPM00.BITNET>
- Date: Sun, 24 Jan 1993 13:35:55 GMT
- Lines: 46
-
-
- This is an important thread (DB2 catalog recovery). Consider the
- goals of the recovery process. These may not all apply:
-
- 1. Restarting DB2 after a failure of DASD containing the
- catalog or directory.
-
- 2. Restarting DB2 after DB2 software failure or operator
- error destroys the catalog.
-
- 3. "Oops, what do we do now" type procedures, e.g. accidental
- dropping of a production database.
-
- 4. Audit concerns, e.g. "who had access to table X last Thursday.
-
- Each of these issues has a different recovery strategy. (1) Generally
- involves a combination of DB2 backup (COPY) and external DASD dumps.
- (2) Is assisted tremendously by external dumps of DB2 data sets. (3)
- and (4) can be further assisted by keeping human readable copies of
- catalog information (such as DBID...) either on paper or in the computer.
-
- Don't forget that DB2 requires all of:
-
- - DB2 catalog
-
- - DB2 directory
-
- - BSDS
-
- - All ICF catalogs
-
- - Logs
-
- Many of these can be recovered from the others (e.g. - recovering the BSDS
- from logs and DB2 catalog and directory). In most cases, this is a slow,
- complex, and tedious process; not something that you want to try the first
- time during a crisis, nor something you would prefer to do at all.
-
- In all cases, the recovery strategy is completely dependant on the speed
- that you need to be able to recover. For example, stable production
- environments with very fast recovery requirements tend to rely on warm or
- hot backup sites or at least DASD dumps of the catalog.
-
- --
-
- Aaron Werman awerman@panix.com
-