Results 1 to 8 of 8

Thread: DR solution

  1. #1
    Join Date
    Dec 2008
    Posts
    44

    Unanswered: DR solution

    Hi,
    We are trying to setup DR system. One of the proposed solution is that copy the database files at the SAN level -- Prod database files are in SAN and these files will be copied to DR database server's SAN real time. The question is whether the copied files can be used to bring up the DR database (when prod is down due to disaster). If yes, what are the steps..? Can I use db2mirroring? Thanks.

  2. #2
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    You should probably look at HADR before trying to do a split-mirror or flash copy for disaster recovery.

    Split-mirror or flash copy can be use to clone a database. You will have to do a DB2 write suspend when the copy is created, but I would contact your SAN vendor for detailed help on how to do this for a DB2 database with their SAN solution.
    M. A. Feldman
    IBM Certified DBA on DB2 for Linux, UNIX, and Windows
    IBM Certified DBA on DB2 for z/OS and OS/390

  3. #3
    Join Date
    Dec 2008
    Posts
    44
    Thanks for the reply.

  4. #4
    Join Date
    Jun 2009
    Posts
    272
    Cholan,
    Basically it is a SAN true copy which you might be talking about. We have this DR set up in our environment. We also perform DR testing.

    Basically at the SAN level, all your data is copied bit by bit. So when you need to bring up the DR environment. You System Admin breaks the mirror [true copying is disabled]. And your filesystems are mounted on DR box. you need to change the host name in db2nodes.cfg to actual DR host name since it will be reflecting your production host name. You can then bring up the instance

  5. #5
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    Quote Originally Posted by blazer789 View Post
    Cholan,
    Basically it is a SAN true copy which you might be talking about. We have this DR set up in our environment. We also perform DR testing.

    Basically at the SAN level, all your data is copied bit by bit. So when you need to bring up the DR environment. You System Admin breaks the mirror [true copying is disabled]. And your filesystems are mounted on DR box. you need to change the host name in db2nodes.cfg to actual DR host name since it will be reflecting your production host name. You can then bring up the instance
    What about rollforward recovery? Do you account for data that is in memory and not written to disk?
    M. A. Feldman
    IBM Certified DBA on DB2 for Linux, UNIX, and Windows
    IBM Certified DBA on DB2 for z/OS and OS/390

  6. #6
    Join Date
    Jun 2009
    Posts
    272
    You are right that is lost. We totally have 4 DR/HA environments since it is the most critical database.

    One is a SAN true copy like I mentioned above.
    The other one is a Q-Replicated copy.
    Q-replicated database inturn has another SAN true copy
    The third one is log-shipping copy.

    And definitely in SAN true copied DR environment, data in the memory is lost

  7. #7
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    Quote Originally Posted by blazer789 View Post
    You are right that is lost. We totally have 4 DR/HA environments since it is the most critical database.

    One is a SAN true copy like I mentioned above.
    The other one is a Q-Replicated copy.
    Q-replicated database inturn has another SAN true copy
    The third one is log-shipping copy.

    And definitely in SAN true copied DR environment, data in the memory is lost
    I am surprised that you don't use HADR. The DB2 license cost for the HADR Standby server is only 100 PVU's (euqivilent to one CPU core), regardless of the nubmer of CPU cores there are installed on the standby. Data integrity iwth HADR is virtually assured.
    M. A. Feldman
    IBM Certified DBA on DB2 for Linux, UNIX, and Windows
    IBM Certified DBA on DB2 for z/OS and OS/390

  8. #8
    Join Date
    Jun 2009
    Posts
    272
    Although now we can read HADR standby database, the time we set up our DR environment HADR didn't have the functionality of read on standby.

    SAN true copy is a cold standby -> License cost is zero for this.
    Log shipping -> this functions based on the script we wrote -> This is still a warm standby and License cost for this is only 100 PVU's.
    Q-Replication -> Here we have the ability to read on standby database. This is used for generating reports. We also write in to some tables in this environment. In HADR we will not have any options to write. Since we anyways have Q-replicated copy, we didn't consider using HADR again.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •