My company is about to star our migration from Kaseya 5 to K2. We've decided that we'd set up a new server, move a couple of clients at a time to make sure we have a handle on how things work, and migrate other clients over slowly.
Our plan is to run K2 within a virtual server, and I had a couple of questions.
1) What should be allocated for system resources for K2? We have around 1300 endpoints, but I would like to provide enough resources that if we expanded to even 5000 endpoints, we'd be ok. We are not currently using BUDR, or KES (though I'm willing to give KES a shot in the future.). Currently we're doing just auditing, monitoring, patching of windows and nightly clean up scripts.
2) What is the recommended (or generally accepted superior) virtualization platform? I'm hearing often that VMWare and Microsoft Virtual Server are similar on paper in basic functionality, but VS tends to choke long before VMWare. Does anyone use Kaseya in a virtual environment who might have some input?
Above is a link to the Kaseya reccomended requirements. Make sure to review if you have not already, as it is a good base line.
I would definately go with VMware Esx(i). This is what we use (3 host clustered enviroment) and performance is pretty good. We have a Kaseya front end and seperate SQL back end. We followed the above requirements, and currently manage about 2000 machines.
Our front end server has 2 vcpu and 8gb of Ram. Page file is on a dedicated partition, not the C:. The SQL back end has 4vcpu and 24gb of Ram allocated. Database, log files, temp database, page file, and DB backup are all seperate partitions. Both servers are running Server 2008 R2 x64 with SQL 2008 as well.
Following the reccomended requirements, all we had to increase was the dedicated amount of RAM. Our entire database is capable of running entirely in physical Ram if needed. Having the fail over enviroment is very nice, but I know this is expensive for some to implement. If running a single host, make sure to have the additional resources just in case you would end up needing more Ram dedicated to SQL.
My company is running K2 on a HyperVR2 platform supporting about 2400 nodes and expanding every day. We have 2GB RAM dedicated to the frontend, and 12GB RAM dedicated to the SQL. We are using KES and BUDR. I will say that only about half of the 2400 nodes are really "in use" and the others basically sit (except for check-in's and one-time patch scans and audits).
I have another VSA that runs on a HyperVR2 cluster (2 nodes) and supports about 600 nodes. All of the nodes are heavily "in-use". About 100 of them use KES and none of them use BUDR. The front-end has 2GB RAM and the SQL has 8GB RAM. It's using all 8GB on the backend, and it's sluggish compared to the non-clustered VSA, but it's still very usable. The plan is to dedicate 4GB more RAM, but the real problem is a disk bottleneck on the SAN that the cluster is sharing.
All-in-all, it's all M$, and highly reliable and performing well.