jetole

With what was suggested, yes you would only need to install it once however you would have a single point of failure since, if your file share goes down then so do all of your web sites and the load balancer would be pointless at this point. If you don’t mind a single point of failure and you are only doing this for faster access / load times then yes, this should increase your load time. One way to do this to avoid the single point of failure would be to either have replication between the two drives which can be done (in Linux and most *nix) with DRBD, a regular scheduled rsync or I am sure there are other ways. Another way would be to have two file servers that sync via DRBD and then use fail over with the file servers. You can do this by using NFS like what volomike said but have a second NFS server with an identical share (via DRBD for example) and if the first file server fails, you can have the second file server take over via use of keepalived or something similar. If you do want to get rid of all scenarios where a single host can cause the site to be down then you would also want to implement MySQL clustering and use some sort of Disaster Recovery (DR) / High Availability (HA) with the load balancer itself. You can also use keepalived for Linux based load balancers i.e. if you are using IPVS (or even haproxy) then you have two identical hosts with one inactive that monitors the first one via keepalived and if it finds that the first one is down then it will take over. The main idea behind HA is that you should be able to walk into the data center, unplug any server. I mean any server, take it off the rack and throw it out the window and still have your site running in a way that clients will never even know or be able to find out that anything has changed when any server has been removed.