January 12, 2010 at 8:19 am
Hi SQL experts:
I’m working on a project in which we need to store huge amount of data (more than 3TB). For customers we only support simple read-only operations which will be finished in several seconds. Due to the disk space and performance issue, we are considering to partition the data to multiple machines.
I can think out a simple plan. I define a partition function, place data to several machines according to the function, and clients should use the same function to determine which machine to query against. But this plan has some shortcomings:
(1)Clients are close coupled with Server.
(2)The partition function should be carefully designed. Use another function in the future is risky because of (1)
Maybe we can add a web service as a query router in front of these servers, which redirects queries to proper backend SQL Servers. But this adds the complexity a lot and the web service could become a bottleneck.
Do we have a better solution for this scenario?
Thanks in advance for any comments
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply