Clustering clarification needed

[jmair]

I'm getting CPU alerts on a Remote Desktop Server. At times, I have 40 clients connecting to the server that has 24 CPUs. I'm wondering if I buy a second server with 24 CPU and cluster the two, will the cluster act as 48 cpus for the 40 clients, eliminating the large performance drops.

 

I ask as I see the Clustering and Failover looks to be an interchangeable word with Windows 2012. Where one is 'clustering' or sharing tasks as one computer and Failover is more of a backup/redundancy.

 

Any clarification would be helpful with my particular problem. Thank you kindly.

[ForgotMyPassword]

Is the RDS server a VM then? If so, what else is sitting on the same hypervisor? What services are published via the RD Gateway? What do the users generally do when accessing the server?

 

If you remote on or connect yourself, does it FEEL slow? Metrics may be one thing but user experience is another and the CPU warnings *could* be disregarded if they are just spikes and no continuously maxxed out?

 

I'd consider having a couple of (VM) RD servers available, maybe even spinning up another VM as the gateway but that's easy to say when I don't know your setup! What's the storage? I ask because on our old setup (plenty of CPU power) using a HP Blade and SAN, the disk throughput was so low, we were seeing CPU warnings...crazy - sorted out the storage (as a temporary thing before moving to a new setup entirely) and the warnings went away. Gotta love working in IT, right?

[jmair]

Thank you for the reply ForgotMyPassword. It is just one Windows 2012 r2 server with Remote Desktop Services installed. Student processes hang on locally loaded typing software or even our required assessments (chrome will hang). The RAM is always good, but the CPU spikes seam fairly regular. But my underlying question is, does Clustering act as a true resource cluster or just a failover when one of the servers fail.

[ForgotMyPassword]

Ah, I see - well, for our setup (three hypervisors all running multiple VMs, storage on SAN over fibre channel), I know I can throw as many vCPUs as I like towards any given VM but I believe they are only addressing the physical CPUs on the host they actually reside on...I guess I'm potentially confusing the situation by introducing virtualisation into the mix but I believe your assumptions re failover are probably on the money.

 

REALLY daft question but what AV are you running and have you looked at excluding processes that you can deem as safe? It could be that AV is hogging a boatload of resources trying to scan the environment for each connected user?

 

Cheers,

James

[jmair]

REALLY daft question but what AV are you running and have you looked at excluding processes that you can deem as safe? It could be that AV is hogging a boatload of resources trying to scan the environment for each connected user?

 

Cheers,

James

That is a really good point. It does generate an instance for each user and of course scans everything altered. I'll definitely contact Sophos about this. Thanks again for your input.