[Gluster-users] HA strategy

[Keith Freedman]

At 01:55 AM 8/14/2008, Rob S wrote:
>Hi,
>
>I'm looking to gluster to help improve the performance of a
>multiserver web platform.  We've a small amount of data (~50-100MB)
>that needs to be shared between all web front-ends.  At present, the
>data is stored in a MySQL blob and replicated out to each front end.
>I've read the documentation at
>http://www.gluster.org/docs/index.php/High-availability_storage_using_server-side_AFR.
>  My question is: Should I run 2 gluster servers, and several gluster
>clients, or could I make each machine a gluster server, and sync using
>AFR?
>
>How are people using gluster?

I have 2 node webserver clusters.  the pair afr eachother.  so it's 
basically what you want.

make sure your setting the local read volume to the local disk and 
that'll help your performance.
I think the only value of separating the servers (and using local 
cache translator) would be that when a file is requested on the 
client, it makes one check to a single server, whereas if you have, 
say a 4 node afr configuration where all the clients are servers to 
eachother, you make 3 checks..  if they all happen at the same speed 
it'll be negligible

that'd be fine if you want 4 copies of your data.

>Rob
>
>_______________________________________________
>Gluster-users mailing list
>Gluster-users at gluster.org
>http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users