Newbie to replication

  • Hello,

    I've been asked to replicate my databases (about 30G) to an offsite location for the sole purpose of recovery if there is a catastrophe at our primary location. The requirement allows for a maximum data loss of 8 hours.

    My research has led me to believe that the best way to do this is with merge replication. This is due to the belief that transaction replication could affect performance and requires all tables to have primary keys or unique indexes and that snapshot replication would require copying and shipping the entire database every 8 hours.

    I'm just looking for advice on my information above, any useful user experiences and any suggestions for tool/utilities that may make this easier, safer or faster.

    Thanks

  • I am unconvinced regarding the use of merge replication for this task. Usually transactional replication is a better way to go (but for performance, don't put the distributor on the same machine as the publisher).

    However your scenario sounds more suited to log shipping (either Microsoft's or roll-your-own).

    HTH

    David Saville

    Aldex Software

    http://www.aldex.co.uk

  • Log Shipping is your best bet if the server taking the updates (Subscriber) will be off-line and will solely be used as a warm backup server in the event of a failure of your production unit. Here is a step-bystep on log shipping:

    http://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/logship1.mspx

    And some further MS documentation:

    http://www.microsoft.com/resources/documentation/sql/2000/all/reskit/en-us/part4/c1361.mspx

    If you need to have the subscriber online (such as users running queries, or data mining), then merge or transactional replication is for you.  If you haven't seen this already, here are the differences:

    http://www.microsoft.com/sql/evaluation/features/replication.asp

    We experienced a similar issue when we set up trans repl for our environment, but we just added a unique ID field to each table that we solely use for repl that didn't have a unique ID to start with.  Worked smooth and has created no problems what so ever.  From your description of what you are trying to do, it sounds like a one-way replication - Distributor to Subscriber; Merge repl can get messy so don't use it unless you really need to, and it doesn't look like you need to here.

     

    Tim

Viewing 3 posts - 1 through 2 (of 2 total)

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