server failure and resuming service

[danIT]

Today our VLE went down, the databse is stored on one server, the data files on another and the actually software/scripts on a webserver, the webserver stopped talking to the data server for some reason, its been resolved. can anyone here think of a way of making this instantly recovrable? could we make a virtual server to kick it to action when a server fials for some reason?

 

any ideas welcome?

[Ric_]

A second webserver and heartbeat to swap over if one fails.

 

Database and filestore relience might be a bit more tricky due to the dynamic nature of their content.

[Geoff]

Yep, you need to double everything up and cluster it.

[projector1]

is clustering expensive though?

[CyberNerd]

Virtualisation would help with hardware failure - if three servers were hosts for 3 virtual machines, then if any of the servers failed the virtual machine would move to one of the other servers. You could add extra virtual servers to this arrangement and have clustered virtual servers. - adding new hardwere when needed.

[Ric_]

@projector1: I'm not well up on clustering but I don't think that it has to be too expenive... obviously hardware costs are going to be at least doubled in this instance.

[Geoff]

Rule of thumb is that the more bomb proof you want it, the more expensive it gets.