November 8, 2009 at 12:56 pm
So one of the clients has a pretty high concurrency environment, around 450 accesses to the web application that access the DB for pretty much everything, and 4-5 users using an application that do the heavy lifting duties importing data. All in all it does well, but its only a matter of time before things begin to derail and we want to split the load on the server in a way that would also serve as a high availability plan.
My first thought on this was to implement peer-to-peer transactional replication on the main tables that both sides of the system use
The importing app imports data to "intermediary" tables, does that checks, transformations and information gathering that it has to do there then sends the data to the "primary" tables that are then used from the web app, so the plan is to only replicate those primary tables to avoid over-stressing the server during data imports.
My concern right now is to the relational information.
The client has a windows server 2003 with sql server 2005 enterprise, he said he doesnt opose to migrating to 2008 if it helps accomplishing the goal.
I would like to hear from you good people, if anyone have any tips or advices, or if ever done anything like this and if it worked out.
Any help is appreciated
Thiago Dantas
--
Thiago Dantas
@DantHimself
November 9, 2009 at 12:45 pm
SQL Server 2008 with Windows 2008 will make a difference because the 2008 OS TCP Stack is much improved. Now the question for you is why Peer to peer? Cant you just have the subscriber as read only?
-Roy
November 9, 2009 at 1:37 pm
no because the subscriber (second server) would be doing the data importing and that data would need to be relayed back to the main server
--
Thiago Dantas
@DantHimself
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply