Setting up Transactional Replication

  • I have never used replication before and I have to set up transactional replication for a number of db's (5) that vary in size (Max 39GB).

    Is it best to have a Publication/Article 1-2-1 or many-2-1 relationship? I was thinking about having 1 publication for each db.

    When setting up replication for the first time can I sync my subscriber by doing a backup/restore from publisher to subscriber and then apply replication? If so, how does SQL Server know if everything is in sync once it starts replicating?

    If a table hasn't got a PK can I simply choose any column within the table to apply the PK to?

    I have discovered in Dev that replication already exists. Can I create a new distribution database to act as the distributor for my replication or is the current distribution db used for all existing and new replicated tables?

    Any help, ideas, suggestions would be greatly appreciated.

     

     

     

  • Hi, I hope to answer your question. First keep it simple. Create a publication for each database you wish to publish. Keep it simple and use the existing distribution db. You may specify a different distribution if you want to. Create a subscription for each publication you wish to subscribe to. Once created the subscription will need to be initialized and this is done by generating a snapshot from the publisher. Backing up and restoring db's across publisher/subscriber is  not the way to sync. The Distribution DB is a key element in knowing that publisher and subscriber are in sync. Replication uses key data columns ie xact_seqno to know the last transaction replicated etc. BTW, you should consider running the Distribution Agent on the Subscribers to take the load of the Publishing server. SQL Server 2000 Transactional & Snapshot Replication by Hillary Cotter is an excellent resource. Hope this helps and hasn't confused you further. Rgds Derek

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

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