DB2 Storage

DB2 system files

These naming standards apply to DB2 running on a z/OS mainframe.
You should have gathered from the previous page that DB2 has loads of files, and many of them are interrelated. DB2 needs to keep track of them, and ensure they are always consistent. DB2 has a number of internal system databases to help do this.

The Catalog Database DSNDB06

contains DB2 system tables, data definitions, statistics, security and recovery information. This is not an ICF catalog, its just another DB2 database. You get information from it using SQL queries, typically to find backup information or maybe inconsistencies between the BOOTSTRAP and DSNZPARM parameters that affect any DB2 recovery.
Table SYSIBM.SYSCOPY, stored in DSNDB06.SYSCOPY, contains information needed by DB2 to manage recoveries, specifically image copy information, and also quiesce points.

The Directory Database DSNDB01

contains similar information, specifically
Table SYSIBM.SYSLGRNX, which identifies log records for updates made to table spaces. Used to select specific log records, and so speed up recovery.

The Work Database DSNDB07

Used for workspace (that was easy!). At least 2 tablespaces will be defined, and they will be very busy, so they are good candidates for isolating on their own disks. If you are managing DB2 backups, then don't worry about backing up DSNDB07.

DB2 Recovery datasets

Bootstrap datasets BSDS

DB2 uses the BSDS to recover from error situations. It is updated every time DB2 issues a checkpoint, and every time DB2 allocates an archive log. The bootstrap dataset is a VSAM KSDS, and its availability is absolutely critical, DB2 won't start up, without 2 of them! They are called
hlq.BSDS01 and BSDS02
where hlq is site dependent

Active logs

DB2 can have up to 31 active log datasets (up to 93 with DB2 version 8). These record all updates to DB2 data, and are used for recovery and rollforward. When an active log fills, it is copied off to an archive log, then the active log is overwritten again. There is some delay built into this process, so some log data will be held on both active and archived logs. Generally, archived logs are held on slower medium than active logs. The overall size of the active log set, must be enough to cater for fast recovery in a reasonable time frame. Active logs are called
hlq.LOGCOPYn.DSxx
where xx is 01-31, and n is 1 or 2 for active log duplexing.

Archive logs

We already know all about these, almost. Archive logs are physical sequential datasets, and can be held on either disk or tape. DB2 will track up to 10,000 archive logs and the naming standard is
hlq.ARCHLOGn.Dyyyyddd.Thhmmsst.Afileseq (or Bfileseq for a BSDS backup)

back to top

Copyright © Lascon Storage Ltd. 2000 to present date. By entering and using this site, you accept the conditions and limitations of use