Skip to Content
avatar image
Former Member

Migrating database from big endian platform to little endian

The document says that index rebuild is necessary but is there alternative to rebuilding indexes or faster way to build them when migrating from Solaris big endian to Linux x86 (red hat) little endian platform.

Appreciate your response..

Thanks

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

2 Answers

  • Oct 27, 2015 at 09:39 PM

    You could run sp_post_xpload, but since that runs serially it will (obviously) take the longest.

    As for ways to speed up the index rebuilds ...

    - look at the source for sp_post_xpload and write your own process to do the same thing but with multiple spids concurrently rebuilding (different) indexes.

    - see the syntax for the 'create index' command, taking into consideration those options that utilize (parallel) worker threads

    NOTE; Keep in mind that rebuilding a clustered index will automatically trigger a rebuild of the non-clustered indexes; net result is you want to make sure you don't rebuild your non-clustered indexes twice.

    Obviously (?) the number of concurrent index rebuilds will be dependent on the your disk subsystem throughput, number of cpus available to ASE, the level of parallelism (if using parallel/worker threads), etc.

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member

      Mark,

      Thanks for the reply. I think you are right we will need to index rebuild in parallel to save some time.

      This is the last thing I want to do but looks like I have no choice.

  • avatar image
    Former Member
    Oct 29, 2015 at 12:27 PM

    Hi Girish,

    On a separate but related note:

    I may need to take a look at a similar migration project - would you mind pointing me to the documentation you have found on this please?

    What has been your experience of this process?

    Kind regards,

    G.

    Add comment
    10|10000 characters needed characters exceeded