Sanity check for RDS deployment + Outlook question
Posted by Flashy-Distance-3329@reddit | sysadmin | View on Reddit | 5 comments
Need to get some sanity check here on the following configuration but I also have a question which will be listed in the bottom:
Infrastructure: 2 Hyper-V servers connected with a 10G connection.
HV1:
- Domain Controller/Connection Broker/File Server
- DB Server (using a solution that does not have the ability to do HA)
- TS1
HV2:
- Backup Domain Controller/Backup Connection Broker/File Server
- TS2
HV Replications:
- DB on HV1 to HV2
- TS1 on HV1 to HV2
- TS2 on HV2 to HV1
Users will be connecting to the connection broker which will then route them to one of the terminals. Folder redirection is set on both terminals to redirect them to a DFSN+DFSR based share so the data is both replicated and always available.
Replications will help me make sure all terminal servers and DB can be available with minimal downtime. I'm not replicating the domain controllers since they will hold the same data thanks to DFSR.
Would appreciate if people here can check this out and tell me if what I'm planning here is accurate and makes sense.
A problem that I have and have yet to figure out is about Outlook configuration since what will happen now is the users will have to create their Outlook profile from scratch on both servers which is an inconvenience. FSLogix does not seem to be the right way to do this due to 0 redundancy options (HA) available on a local environment. I was proposed forcing OSTs to be created on the DFS share and incorporating roaming profiles, but it sounds like that might be the wrong move. I've never seen OSTs running off a network drive, let alone on a DFS. Wondering if anyone can recommend for a better way of accomplishing it.
Thank you all for looking at this in the first place!
sembee2@reddit
I don't see the point in replicating terminal servers. Is there something different about them? If not, put thrm in to a single farm, 100% identical with just the name being different. Then use FXLogix. I would usually folder redirect Docs, Pictures etc out.
Flashy-Distance-3329@reddit (OP)
Thank you for your reply! The reason we were wanting to replicate the terminal servers is to have redundancy both in OS and Hardware level. From our experience, having too many users on the same terminal server OS even though they have enough resources would experience degradation in their experience. Therefore, while having an outage on one of the servers and everyone will divert to the second server will make things available, If we need to service the original server for a long time they will end up suffering on a single TS.
The problem I understand I will encounter with FSLogix is I can't save the data both on FS1 and FS2 that way if one of them is down, the other one is still available to service the users. Am I missing anything about how this works?
man__i__love__frogs@reddit
If you don't want shared storage like a SAN, you would use something like S2D to join the storage from the 2 hyper-v's and then set up HA windows file server.
You could also just have DR replication setup and restore to the most recent one, is someone's appdata being x hours out of date going to impact their work?
CP_Money@reddit
Not using FSLogix for this is such a bad idea. I have been doing RDS since Server 2003 and it is the best thing to ever happen to RDS.
Scottehhhhh@reddit
Failover cluster with shared disks through some sort of hyperconverged virtual SAN, then FSLogix stored on a network drive.
That provides your redundancy for the OS/hardware (i.e. loss of single host = VMs survive on remaining host, running on the mirrored disks from that host), and allows your profiles to sync between both RDS servers in the farm.