Infrastructure scaling assessment and recommendations

This guide covers one way to get an idea of average emails sent per minute. This information can be used to determine if the default implementation cluster configuration for Templafy's Email Signature Server is adequate for your organization, or if more nodes and pods may be required.

Terminology

Kubernetes uses its own terminology for underlying technologies:

• Cluster: This is nearly synonymous with Kubernetes itself. This is the collection of Nodes and Pods that comprise both the applications which deliver a given solution as well as the processes that make up Kubernetes and allow it to automatically maintain and manage the solution (server).
• Node: The actual virtual machines running in your Azure subscription that will host Pods. These are what have actual costs, billed by Microsoft, associated with them running in your organizations Azure subscription.
• Pod: These are docker containers that are running a single instance of a Templafy Email Signature Server. Pods run within a Node and are distributed automatically across the Nodes in the cluster by Kubernetes.

Review the Email Activity Reports

1. Go to https://admin.microsoft.com/Adminportal/Home#/reportsUsage/EmailActivity.
2. Review the various peak Sent email points for 7, 30, 90, and 180 days.
3. Look for trends and the highest peaks for the Sent emails data point.
4. Divide this peak number of emails sent per day by 8, 12, or 24. Depending on what best fits the working hours of your users.
5. Divide this peak number of emails sent per hour by 60.
6. This number will provide a peak load of emails sent per minute. Use this figure to consult the table below for guidance on what should be the most appropriate initial implementation sizing.

Specifications

• Pricing can vary depending on the Azure region.
• By default, Kubernetes Node and Pod auto-scaling is enabled. This can, at maximum, double the number of Nodes, Pods and Emails/minute processed if the load on the cluster requires.
• The current default implementation is 3 nodes / 34 pods.
  • Pod CPU range: 300-500 Mi CPU (1000 Mi = 1 vCPU)
  • Pod Memory range: 300-800 MB
  • Ratio pods per node is dependent on the Virtual machine type chosen.
    • DS2_v2: recommended 3 pods per node.
    • F4s_v2: recommended 7-10 pods per node.
  • Avoid situations in which all your pods are full in a node.

Client-Hosted Infrastructure recommendations

Baseline: 25 emails/pod (average taking into account different mail types and sizes)