Kaseya Community

"Has anyone hosted their KServer on AWS (Amazon Web Services)?" and other I/O latency questions...

  • We're running a ~100 GB database with ~7,300 clients on NetApp SAS storage; however, our disk latency is past the 25ms timing which causes issues with any MSSQL Database (not just Kaseya's instance) per my understanding when talking to Kaseya's performance gurus:

    I have a few questions for the community:

    1) SaaS uses flash storage for their databases (both ksubscribers and TempDB) and have thousands (if not over 10,000) endpoints per SaaS server, per their Performance TechJam: http://community.kaseya.com/resources/m/techjams/74666.aspx   What are other folks doing with their on-prem servers?  How many clients have you got, and are you running on fast disk (i.e. SSDs)?

    2) How are y'alls disk latency numbers (assuming you're virtualized, what does the screen above look like)?

    3) Has anyone here successfully run a high-volume KServer (frontend, backend or both) in Amazon Web Services?

    Just looking for some more data points before we pony up for SSDs, move to AWS, or...?

  • 5Gb database, 200 agents. Max latency 1ms. Average 0ms. HP DL380 on ESXi 5.0 with internal SAS 10k drives and a P411 caching RAID controller.

    Useless numbers from your perspective, but never the less, at least I replied :)

    Is your SAS array direct connected, FC or iSCSI? An switches involved? There can be noticeable latencies in lower end gear such as iSCSI over a gigabit port or two with a store-and-forward based ethernet switch in there somewhere. A direct-connected FC array will have a much lower latency, by comparison.

    The point being: A re-organization of your existing infrastructure such as removing a slow switch may solve your issue....