[Gluster-users] slave as a failover server using Geo-Replication

[Andre Mosin]

Hello,

I have a high traffic Wordpress website which is hosted in a cluster environment. All the content of the site (about 40GB) is hosted on a dedicated NFS server. I have a dedicated gigabit switch for the communication between the servers. I was looking a way to have more redundancy content and a possible failover systema and Gluster is the exact solution that I was looking for. This is a great project but I have tried using gluster on my environment with the replica option (2 servers mirrored) and I noticed that the backend of Wordpress hangs and it gets too slow. I am not sure what could be causing this problem. Maybe a gigabit switch is not enough? or my gluster settings are not correct for this type of scenario... I tried mounting the clients using glusterfs and NFS and both ways are slow.  With my topology would I be able to use gluster and not loose too much performance? I am not sure what Wordpress does when we open the back-end but I just notice that problem on the back-end of the site.. the front-end looks good.

Well if not possible I guess I might be to use the new Gluster Feature to get more performance? My questing here is... Can I use Geo-Replication slave not as a back up system but as a Fail over server?  I can set up Linux HA for example to do the fail-over system.


Regards,

Dantas
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20110627/b57ef79f/attachment.html>