April 23, 2007 at 9:17 am
In SQLServer 2005 Standard addition using transaction replication if the publisher fails can the subscriber be changed to a publisher. If so, how is that accomplished.
The configuration we are using consists of 3 regions (for geographhic failover), with each region containing two database servers (one active, one stand-by) with an external RAID array containing all database files using mirroring to failover within a region to the alternate server.
One region will be a publisher with the other two being subscribers. In a catastrophic failover of the publisher region, we need to bring up a subscriber region as a publisher within hours. We are aware of the Peer to Peer feature but that is only in the Enterprise version which unfortunately is cost prohibitive. So, is it possible to change a subscriber to a publisher, and if so what would be the procedure to do this.
April 24, 2007 at 10:37 pm
What type of replication are you using? With Merge Replication you can setup a publisher to become a distributor.
Catherine
Catherine Eibner
cybner.com.au
April 25, 2007 at 6:56 am
We're using transaction replication with log shipping (not merge replication) and are trying to figure out how we would make a subscriber a publisher in the case of a catastrophic failure where the region with the publisher is completely gone.
May 2, 2007 at 4:44 pm
When you configure for repl [to be a pub] you designate the Dist too [usually same box] which adds a fistfull of sytem tables and the distribution db.
You do not have to configure a sub since this is usually passive [DistribAgents usually PUSH from Dist], but exception is if you want PULL instead [hence would need config to get DAs].
Apart from the classic Pub-Dist-Sub model, you could cascade, ie A-B then B-C, B-D etc
- might be appropriate for London-Singapore then Singapore-Tokyo etc
Hence B is both sub [unaware whether changes are by local client or remotely from DA on Dist] and pub.
In your case you would pre-configure [in case of emergency] but not have any active subs
- read up on "Not For Replication" attributes
recognise there are plenty of alternatives like Clustering (virtual server with failover) to cope with failure/outages. And Spanning-Tree to protect you from individual network segment failures. You should consider you datacentres, links, MTBF, MTTR and business SLAs.
enjoy !
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply