We're looking to replace a single, heavily used 2003 TS server--for performance, feature, and capacity reasons. Our users frequently have small amounts of data on their desktops/settings that we'd to persist between sessions, and they tend to have
long-running sessions (disconnecting and reconnecting while traveling, but not logging off--keeping open apps in the meantime).
The servers were originally purchased intending for 2008R2 SP1 remote desktops services, with lots of fast internal (RAID 5, 12x300GB 15K) storage. We're considering the option of using 2012 (virtual sessions, not virtual desktops) so that we can scale out as we grow--and use the User Profile Disks.
Ideally, we'd like to maximize the usefulness of the purchased servers (and their internal storage), and not have to purchase additional hardware for shared storage--(iSCSI/External array that can be clustered). We're wondering if it's possible to pair the User Profile Disks (UPD) with DFS replication (possibly over a dedicated NIC). Then, a user could log in to server X, connect to her local UPD (with the changes replicating to server Y's copy). If she disconnected/reconnected, the RD connection broker would connect her to back to her existing session, and if she logs off and back on, she could connect to either server X or Y and it would all work. For maintenance, we'd be able to drainstop one server via the connection broker, perform the maintenance, let DFS catch up, and then do the same on the second server.
Would something like this be possible? Or is it just asking for major problems?
The servers were originally purchased intending for 2008R2 SP1 remote desktops services, with lots of fast internal (RAID 5, 12x300GB 15K) storage. We're considering the option of using 2012 (virtual sessions, not virtual desktops) so that we can scale out as we grow--and use the User Profile Disks.
Ideally, we'd like to maximize the usefulness of the purchased servers (and their internal storage), and not have to purchase additional hardware for shared storage--(iSCSI/External array that can be clustered). We're wondering if it's possible to pair the User Profile Disks (UPD) with DFS replication (possibly over a dedicated NIC). Then, a user could log in to server X, connect to her local UPD (with the changes replicating to server Y's copy). If she disconnected/reconnected, the RD connection broker would connect her to back to her existing session, and if she logs off and back on, she could connect to either server X or Y and it would all work. For maintenance, we'd be able to drainstop one server via the connection broker, perform the maintenance, let DFS catch up, and then do the same on the second server.
Would something like this be possible? Or is it just asking for major problems?