Hypothetical Server Setup


Recommended Posts

I am a fish out of water on this one, but am going to ask since I am working on a research project for my Uni Class. Basically, I have been given the role of being the manger of a Network Architecture Department ( I can set my own network up at home like everyone else and do some basic stuff, but nothing near what a corporation would) I have no formal training in this area.

The task is to write up a recommendation for the purchase of servers (if needed), hardware etc to allow 75 more workstations. There are already 250 employees presumably on one or two servers. (They didn't specify the current setup in the assignment)

One thing to note - I am my first term and this is a Communications class no official training in this area. I am majoring in Information Assurance and Security which I am sure this will come around later on in the next couple of terms. But for now, things are just hypothetical and they don't expect you to know specifics.

In the assignment, I am assuming that the server(s) is running fine in this scenario but would this be a bottleneck already if they are one one server? If two, load balancing would work I presume, but If they are adding 75 additional users, would an additional server be recommended?

If so, what type of hardware would I want to start looking at? I am centering somewhat on the Dell Poweredge C Series, but might be overkill? Not sure.

Not looking for answers as I must cite references and sources to vendors etc, but any direction in this area would be a great help here.

Link to comment
Share on other sites

Having some additional information about the environment would be useful. Mostly it just depends on the workload the server is currently handling and what type of work it is. Virtualization is a wonderful thing for most workloads, so building around that helps. It also depends on what the projected growth of the company is in the next 3-5 years (the average life of a tier 1 production server).

Just running with a few guesses is that the server is nearing its end of life and disk usage is at say 75% (400gb storage, 100gb avaliable), CPU usage is averaging 60%, memory is bottlenecked due to 32bit OS. Lets say its running SQL Server 2005 standard edition 32bit.

I would consider just the regular Poweredge R720 or R820 with Server 2012 and Hyper-V, build two with SSD caching and spinning disk for storage. Expandability on those servers is pretty good, the R720XD are really nice for huge amounts of storage (i have one with 40tb avaliable in raid 50). Setup Hyper-V with replication to the second identical server and your recovery time is under 5 minutes and your at most 10 minutes out of sync. Stick with 1g or 10g networking for the servers (10g is always wonderful trust me but costs a fair bit extra, i am going to assume money is am object though so 4 teamed 1g ports on the server is very acceptable most of the time for that many clients) and 1g for backup only so you do not step on production bandwidth.

Failover is easy with that and updates for the hosts is pretty painless. Update and reboot your current destination host and test it, then force a failover. It will sync and shutdown the guest and bring it up on the other machine. Do this for all of the virtuals and then bring down the now destination host for updates. Pretty simple and depending on the guest requirements its not going to be that expensive.

Clustering is always an option but most of the time just adds a lot of complications. you have to spend the money on shared storage, your expandability is nice but again the life of most production servers like that are 3-5 years max. If you can expand the bottlenecks such as disk and memory throughout the life then you are normally ok.

Hope it helps

Link to comment
Share on other sites

This topic is now closed to further replies.