Thoughts and Opinions on our new network setup
Hi folks, yet again another question :D
SCCM 2012, working very well and very happy with it's capabilities.
We wish to add a bit of resiliance/failover and convenience and DFS seems a good way to acheive this. We are a split site school connected with a gig fibre, and yesterday we lost one of our servers over the other site. So all users on that server had a fair amount of downtime.
I believe with the use of DFS, both namespace and replication we can negate that sort of thing entirely - keeping all users across both site servers, if one dissapears then they lose nothing, can still log on thanks to the other server etc. However we've been a little bit perturbed by some of the stories on Edugeek, some people enjoy success with DFS, others do not. I can't for the life of me figure out a better way of doing it though. Users will be kept locally on the servers, we could at a push knock up a couple of selfbuild SAN's but would really rather not. Not keen on single points of failure (i.e. one server/storage unit with all users), there's plenty of methods of replication but would rather be able to have no downtime for all users rather than downtime for all or downtime for some.
One of the questions that appears to come up is network load. Again, I'm not keen on having separate servers for just about everything as it seems a bit of a waste of resource - so the plan is a server on each site, both are DCs, the main server runs SCCM and SQL too. This I know is pushing it a bit but also thinking about separating some parts.
Any input on this sort of thing? It looks fairly simple to set up, it'll only be users that are replicated. Servers are self build dual xeon 5620 and dual e5-2620s, 16-32gb ram and enough SATA storage to shack a stick at. ESXi but mostly for snapshot/veeam backup ability (they won't be running anything else unless it would help at all to separate some services)