User Profile Discs



  • I'm looking for recommendations for User Profile Disks. I currently Have 4 RDS servers (host-based, not VDI) in a farm and have previously implemented User Profile Disks. UPD's worked great all being served up from a standard VM file server, until our user base grew to well over 100 people logging in every morning all at once. What's happening is every morning between 8-9 AM, there are so many people loading up their profiles that everything essentially freezes on all of the RDP hosts until the VM serving UPD's can catch up.

    I've been researching best practices for load balancing for our user profile discs and have seen some saying that Scale Out File Server is the way to go. Does anyone have any thoughts or experience in a situation like this? I want to make sure I have this set up in such a way that it can scale further in the future if I need it.



  • Have you place a limit to the UPDs? When creating them you can set a limit and once they have been created very few things change.



  • @dbeato are you referring to a quota on the size of the UPD's? I apologise if I'm not understanding what you're referring to. If you are talking about quotas, yes, they're limited to 20GB/ea.



  • @dmacf10 said in User Profile Discs:

    @dbeato are you referring to a quota on the size of the UPD's? I apologise if I'm not understanding what you're referring to. If you are talking about quotas, yes, they're limited to 20GB/ea.

    Yes, quotas on the size of the UPDs.



  • It seems to me the issue he's having is more of a logon storm where 100 people are all trying to logon and connect to their UPDs at the same time.

    Things I would check on...

    • Where is the VM file server in relation to your RDS servers? See if you can get them into the same network switch. (IE: Migrate your File Server into the same Network Rack as your RDS servers).

    • Does the file server that handles the UPDs serve other files too? If so, split them up so that the UPD file server only serves UPDs.

    • How fast is your network to the File Server VM?

    • How fast is your network to the RDS Servers?

    • How fast is your network between switches ?

    I'm unfamiliar with SOFS, but a quick read (https://blogs.technet.microsoft.com/filecab/2013/12/05/to-scale-out-or-not-to-scale-out-that-is-the-question/) makes me think it could work for you...



  • @dmacf10 said in User Profile Discs:

    User Profile Disks

    Before you start moving things around, I'd try to get some monitoring in place to see where your bottleneck is now. If the server's disks just can't keep up it doesn't matter where things are located with regards to networking.

    Given what you've described, it's a reasonable expectation that there might be gains to be made on the network side of things, but don't forget the basics:

    • Host firmware updates
    • Host BIOS settings (SR-IOV and friends)
    • Hypervisor updates
    • Networking
      -- Dedicate a NIC to this VM if possible
      -- Look into 10G networking if not already there
      -- Jumbo frames?
      -- SMB optimization (SMB3 / other optimizations)

    .... I'm all about eeking the most out of what's already there before throwing something new into the mix...



  • @notverypunny Yeah, monitoring first. Where is the bottleneck? Is it CPU, IOPS, RAM, network, all of the above? Which servers are overloaded? Is it one or multiple RDS servers or the file server?



  • The entire network is running 10Gb fiber. All RDP hosts are separated by only one switch. The Network isn't saturated by any means. I'm thinking that it's a issue with the storage on the server that's hosting the UPDs. It's a 4-disk raid 10 with WD 7200 RPM Enterprise drives. I just want to make sure the next move I make will take care of this issue and allow me to scale up a bit more as we add more users.



  • @dmacf10 said in User Profile Discs:

    The entire network is running 10Gb fiber. All RDP hosts are separated by only one switch. The Network isn't saturated by any means. I'm thinking that it's a issue with the storage on the server that's hosting the UPDs. It's a 4-disk raid 10 with WD 7200 RPM Enterprise drives. I just want to make sure the next move I make will take care of this issue and allow me to scale up a bit more as we add more users.

    Uhm, yeah, that's not going to handle the login storms, and would not be very fast in general. I know OEMs charge way to much for SSD, but they're getting cheap for the capacity now. https://www.newegg.com/Product/Product.aspx?Item=9SIA99494V9845 $225 for an Intel DC 1.6TB.



  • @travisdh1 This^^



  • @wrx7m said in User Profile Discs:

    @travisdh1 This^^

    If I ever feel the pinch of having "only" 1.5TB available on my home lab box, I know what I can do for an upgrade!


Log in to reply