maintaining VLDBS - 1 tb

  • I am trying to document best possible way to administer a 1+tb of db. Please try to answer the following questions:

    i) i will be splitting mdfs in 4 different files spread across 4 luns with first three set to 250 gb and last one to unlimited....any other suggestion on this?

    ii) how should i rebuild index for such a huge db?

    iii)i will be having one full backup on monday and consecutively once differential backups every other day and log backups every 15 mins for point in time recovery ...any other suggestions?

    thanks in advance.

  • iqtedar (5/19/2010)


    I am trying to document best possible way to administer a 1+tb of db. Please try to answer the following questions:

    i) i will be splitting mdfs in 4 different files spread across 4 luns with first three set to 250 gb and last one to unlimited....any other suggestion on this?

    This method can work fine. I would take the tables of the database and split them into the various filegroups and files depending on usage. If you can only have one lun that can grow, then place your tables that are most likely to get large in the filegroup that will be on that lun.

    ii) how should i rebuild index for such a huge db?

    If you build your filegroups like I just described then you can rebuild indexes based on filegroups and setup a schedule that alternates through the filegroups so the entire database is not impacted all at once.

    iii)i will be having one full backup on monday and consecutively once differential backups every other day and log backups every 15 mins for point in time recovery ...any other suggestions?

    That should be fine - depending on your business requirements for acceptable data loss and recovery.

    Jason...AKA CirqueDeSQLeil
    _______________________________________________
    I have given a name to my pain...MCM SQL Server, MVP
    SQL RNNR
    Posting Performance Based Questions - Gail Shaw[/url]
    Learn Extended Events

  • I'd agree with Jason. One more thing I'd suggest is that you rebuild indexes only on fragmentation, not on a schedule. There are scripts here on the site to help you do this.

Viewing 3 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply