May 6, 2004 at 11:48 am
Were evaluating using a network switch as a failover "manager". The switch will direct traffic to either of 2 SQL servers depending on the heartbeat. The 2 servers will be constantly connected (unless failure occurs). The 2 servers do not share a common disk, so each will have different database files ... What replication model would work best for this? What latency/locks will occur and has anyone tried this successfully?
May 7, 2004 at 11:06 am
merge. latency, relative to other models, is the lowest. there would be no connection/transaction continuity of course. so, anything in progress on the failed server would be lost. i don't know what "locks" would mean in this context. yes, we ran this type setup at my last gig. the db's serviced matching web servers in a 3X3 setup. two web/db pairs where on the same lan and one web/db pair was over a dedicated t1. the switch handle failover for any web or db failure.
merge replication works great . . . but you have to be meticulous in handling it. it also requires close coordination with the development community.
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply