Hi,
To add to Gordon's document, let me explain how call is handled by CWMS which can provide some clarity.
1. When you dial into CWMS, you use a Route Pattern (Route Group-Route List) that points to Load Balancing Point SIP trunk
2. The call reaches CWMS Load Balancing point which answers the call and plays Welcome prompts
3. Once you enter the meeting access number, Load Balancing Point is checking internal resources to see which Application Point will be hosting the meeting. Once that is identified, Load Balancing Point initiates a SIP REFER and uses a SIP Route Pattern on CUCM to route a call to a specific Application Point via Application Point SIP Trunk.
4. Hence, this referred call goes back to CUCM > SIP Route Pattern > App Point SIP trunk and comes back to CWMS to the Application Point where the call is connected to the meeting.
Now, on 50 user system, Media server is integrated in Admin VM, so your LB and AP are on the Admin VM. Hence,the call will come in to CWMS LB Point via LB SIP trunk and port 5060, the prompt will be played and meeting access number entered, and then will go out to CUCM, use SIP Route Pattern, and come back to CWMS APP point via AP SIP Trunk and port 5062 and connect to the meeting.
On 250/800 user systems, Media VM is the one where Load Balancing and Application points are, so your SIP trunks will point to Media VMs.
Only in 2000 users systems, or in 50/250/800 with High Availability, will you have multiple Application points (multiple Media VMs), so this entire mechanism starts to make sense.
For example, in 250 user system with HA, you will have 2 Media VMs that loadbalance all the time.
Hence, each of this Media VMs will have a Load Balancing Point and Application Point.
So you can have a scenario like this:
1. Call comes in to Media 1 LB Point
2. You enter meeting access number
3. LB Point identifies that this meeting is on Media 2 App point
4. LB Point sends a call via CUCM > SIP route pattern that points to Media 2 App Point SIP trunk, and the call comes to Media 2 App point and the call joins the meeting.
I hope this clarifies why you need to have 1x LB SIP trunk, 1x APP SIP trunk, and 1x SIP Route pattern for each Media VM (or Admin VM in 50 user system) in your deployment.
-Dejan
