webserver cluster serving web content using a SAN still results in a single point of failure

Question

I'm looking for a better way of doing things than using a SAN lun mounted to multiple webservers as a way of keeping data in sync. The applications require instantaneous or near instantaneous synchronization between webservers to function, and can't be modified to work another way. OC2FS on an iscsi SAN is a good way to do it, but it still leaves the lun as a single point of failure. For example, if for some reason filesystem level corruption happens, you need to mount it as read only on all webservers so you can fsck it, which is not acceptable in my application. If you sync it over to a spare lun as a cold spare with say, drdb, which you could switch to manually and fsck the original, you sync the corruption over as well, and run into problems with your data on the original getting out of sync with your backup lun.

Is there a better way to provide some reduction in downtime in the case of problems with the SAN?

Answer 1

if SAN goes down and all of your web server resides on SAN then you pretty much out of other options unless you get yourself a redundant SAN that would work for you in case if first SAN is down.