We've run web servers in primary/backup mode for years, but are now
looking to cluster two servers and potentially add a third. Currently
we run the servers using Unison to sync the data partitions, then the
secondary server takes over web services using its local copy of the
data should the primary fail. This process is fairly slow (unison runs
twice a day examining about 60 Gbytes of data.
I'd like to move toward a clustered environment with both servers
responding to web requests and maintaining a redundant shared data
store. This would allow us to add a third server to the cluster for
increased reliability and performance, while maintaining two nodes as
the redundant "file" servers. The question I have relates to the best
way to share the data between the first two servers.
I'm looking at two potential options for redundant data storage on the
network:
1) DRDB to mirror a data drive from ServerA to ServerB, then use NFS
failover to mount whichever one is acting as primary. Could GFS work
instead of NFS here?
2) NDB + RAID to mirror a data drive from serverA to serverB, then use
NFS failover to mount whichever one is acting as primary. Again, could
GFS work here?
I'm looking for recommendations from anyone who has specifically used
either of these two solutions in terms of ease of management, speed, and
above-all, stability of the platform. Also, we're planning on using ext3
as the filesystem. Any recommendations otherwise?
If I missed anything that might work well in this senario, I'd love a
heads-up. Unfortunately, shared SCSI, and that type of dedicated
hardware etc., is not an option.
Thanks in advance for the help and advice.
sean
--
Sean Fulton
GCN Publishing, Inc.
Internet Design, Development and Consulting For Today's Media Companies
http://www.gcnpublishing.com
(914) 937-2451, x203
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
