I'm completely well-aware of the fact that I can classify myself as insane. To exemplify this, I was stricken with an absolutely insane idea that I may consider following through with in the near future, tested utilized Virtual Machines, however.

A little back story first.

I work in managed hosting as a Systems Administrator. Our environment is made up of a large variety of servers, mixed between Linux, FreeBSD, and Windows. One of our occasional setups are High-Availability, or HA, clusters. In short, the idea is that there are two or more nodes that comprise the cluster, and provide redundancy should one of the nodes go down. On Linux, we accomplish this utilizing DRBD, and a clustering software such as Heartbeat or Pacemaker. The basic premise of DRBD is that data is replicated between the nodes at the block-level, so any node is constantly up-to-date should the other node go down.

What's so insane about this?

This kind of setup is intended for services that need to be highly-available, such as databases, web services, etc. The backend of all this is DRBD, providing a consistent replication of all the data between nodes, and it doesn't care what's stored on it. So in theory, it's possible to make anything highly-available. The insanity comes in with my next project, just to mess around: The HA Workstation!

Yes, I intend to setup a Linux installation, with X, installed into a DRBD cluster. I'm totally insane. There are certainly some potential applications for this, however, if I can get it to work. HA Terminal Server anyone?

Stay tuned for when I eventually start this project, and run up documentation along the way!