I have created a NLB CAS array to help with RPC latency issues. It is working well and I can see clients using the CAS array, but when I created the CAS array in the default site it included my primary HUB T/CAS server as well, this is ok, but the HUB/CAS server is not running NLB only the two CAS array servers. This of course means the HUB/CAS server is not load balancing the majority of the client connections. Is there a way to set an order or weight he auto discover to use the HUB/CAS server as a last resort? I know I could probably just add the HUB/CAS to the NLB cluster, but it would require quite a bit of redesign. Here is a brief overview of my installation. My only goal here is to have the outlook clients connect to the CAS to keep that load off the primary Hub/Transport server.
if the CAS servers are not in NLB then creating the array is currently useless, clients will still connect now to the CAS server name as if you didn't the CAS array, if you create new mailbox database it will use the CAS array FQDN as RPCCLEINTACCESSSERVER.
so the FQDN you created now just points to a single server and thus the failover process is manual.
the CAS array FQDN is just a name that points to IP that should point to a VIP that load balance your CAS Traffic, when you create the CAS array you supply its FQDN.
old mailbox databased will not use the CAS array so you must change it manually, so if you don't use NLB it is just another FQDN that is same as CAS server FQDN so it will not load balance the traffic nor provide HA. the CAS array itself do nothing.
I have created a NLB CAS array to help with RPC latency issues. It is working well and I can see clients using the CAS array, but when I created the CAS array in the default site it included my primary HUB T/CAS server as well, this is ok, but the HUB/CAS server is not running NLB only the two CAS array servers.
No it didn't. When you use Get-ClientAccessArray it'll show you the *possible* candidates for inclusion in the WNLB or Hardware LB, it doesn't actually mean it is doing anything with them. It is up to you to actually put the nodes you want into the load balancing mechanism of your choice.Microsoft Premier Field Engineer, Exchange
MCSA 2000/2003, CCNA
MCITP: Enterprise Messaging Administrator 2010
Former Microsoft MVP, Exchange Server
My posts are provided "AS IS" with no guarantees, no warranties, and they confer no rights.
Yes I only have one transport server at this time. Based my Organization size we should be ok with one. What i experienced when all of my clients start hitting the HUB/Transport/Server is high rpc latency and slow outlook connections. I had a premier case open, and Microsoft stated the answer was to turn on cache mode. After doing this it seemed to fix the issues. But i wanted to create the CAS cluster to ease that load even further on the main CAS/HUB/Transport server, and for the most part the CAS NLB is doing its job I see 150 clients per CAS array server. The problem with removing the CAS role from the main server is that everything is working, and my user base has absolutely 0 patience or tolerance for downtime.
Only the servers that are being load balanced will be in the array.
The problem you may be seeing is that the CASArray is not defined in the RPCCLIENTACCESSSERVER property under the get-mailboxdatabase command.
If you do a " Get-Mailboxdatabase | FL Name,RPCLIENTACCESSSERVER" You will get a list of all the databases and their connection points. You may need to update those databases with the wrong connection point by doing a " Set-Mailboxdatabase -Identify databasename -RPCCLIIENTACCESSSERVER casarray.fqdn.com
For more information on the new RPC features check here:
I assume you have pointed all mailboxes databases to the CAS Array, as you said “I see 150 clients per CAS array server”. Since main server isn"t in the NLB, and the mailbox databases haven"t pointed to it, I think it won"t handle the client request. So, removing CAS role on the main servers shouldn"t be a problem as long as all the services is pointing to the CAS Array
But I"m confused about your words below. As you said, you want to ease the load, why you try to remove the CAS role rather than add the main server into the NLB? Isn"t that the way to ease the load? Please correct me if I have misunderstood the question
Quote: “But i wanted to create the CAS cluster to ease that load even further on the main CAS/HUB/Transport server … The problem with removing the CAS role from the main server is that everything is working, and my user base has absolutely 0 patience or tolerance for downtime”
I have pointed all of the mailboxes to the array, but for some reason when i created the CAS array it automaticly added all of my cas servers to it.I think it was supposed to ,but Mr. Day above disagrees. The concern with removing the CAS roll on the Main CAS/HUB is that OWA and Active Sync and BES and all of that jazz are working, and removing that role will break that. I never intended to have the Main/hub server in the CAS array its was an unindended negative result.
Mr. Day isn"t disagree what you saw. What you saw is the expected behavior. The array object is logical, and will contain all the CAS in an AD site, but has no impact on traffic or load balancing, which is the job of the admin to control by using NLB
About removal of the CAS role, please make sure that all the CAS related services are provided by your other two CAS servers, as you said, like OWA, ActiveSync and BES. Then, you can safely remove the CAS role