Originally Posted by CanadaGradeEh
Thanks for a pretty much dead-on explanation of what I was about to ask, SoBe! I'm going to be starting a little term project for my networking class soon, and that's what I was considering doing.
JUST TO BE CLEAR, though:
So, I have three computers - one load balancer, two nodes to do the workload - and either of the two working nodes can be used as the back-up node incase the other fails or dies, right?
Also, another question, which is basically what dualgeek asked:
Do both working nodes need to be identical hardware-wise?
Does the load balancer need to be higher-end than the working nodes? I'd think it wouldn't have to be high-end at all.
First off, the machines can very quite a bit. It would help with performance if the machines were alike, but it's definately not required.
Second, I didn't really explain the load balancing aspect very well, I apologize. I have never actually set up a load balancing server, so it's hard to explain exactley how it works. In a load balancing cluster, ALL the machines do equal amounts of work, there is just one server that decides what goes where. Not sure how it deals with down servers.
I am not sure however, if you can do both clustering modes at the same time though. It's something you can easily research. It really depends on what kind of setup you want. Do you want redundancy or performance? For data libraries a failover cluster would be best, but a streaming media server would work best with a load balancing server.
What are you going to use as an OS for your servers? Over here, we use windows server 2003 enterprise edition. It has to be enterprise edition for the clustering. That may be a little hard to come by. There are some Linux Apps that provide clustering services. I don't know much about it, but i heard it's pretty easy.