RMS uses Azure Kubernetes Service for the compute services needed for the API, streaming and encoding. When you deploy RMS, you have the option of using the default VM family and size for each node pool or selecting your own. In all cases, you need to make sure that you have enough vCPU quota allocated for the region where you are deploying RMS.

NOTE: adding quota does not add to your costs. Adding quota ensures that AKS can scale when it needs to scale. vCPU not actively used is not billable.

The default VM families are Standard FSv2 for encoding and Standard DASv5 for the system and workload pools. The default VM for the encoder is F8s_v2, which requires eight vCPU per VM. To ensure sufficient vCPU quota for encoding, select at least 40 cores. This will allow five concurrent encoding jobs. You'll want a number that corresponds to the number of concurrent encoding jobs you anticipate. If you want to support 20 concurrent encoding jobs, then add 160 vCPU for the Standard FSv2 family in the region. 

For the system and workload pools, add at least 28 vCPU for the Standard DASv5 family in the region. The default VM is the D2as_v5 that has two vCPU per VM. These pools are dedicated to the API and streaming servers. Again, adding the quota doesn't increase the number of servers running, it just ensures that the system can scale if it needs to.

How to add quota

1. Enter “quota” in the search bar at the top of the Azure portal page and then select Quotas from the list.

2. Select the Compute tile

3. Pick your subscription and region, and then search for Fsv2

4. click on the edit pencil icon and enter the new (higher) amount of vCPU. This doesn’t trigger costs, it just allows that many to be used if needed by any resource in the subscription-region.

5. Do the same for the Standard DASv5 family