Throughput Configuration

How Do I Achieve Increased Throughput?

To enhance the throughput of Constellations, the max replica count of specific services may be increased. This can be achieved by implementing the recommended ratios for each service.

Before increasing the replica count, it is important to calculate the required resources and provision these in Azure. This allows Kubernetes to appropriately allocate the pods.

Determine the desired number of files that need to be processed concurrently by Constellations, the set the CDR Enablers to that number. Based on this value, you may apply the following ratios to adjust the replica count for the relevant services:

• 6 CDR Enabler : 1 Sync API - To effectively handle the requests from every 6 CDR Enablers, it is recommended to have 1 Sync API dedicated to managing the requests of those enablers. This allocation ensures efficient coordination and processing of the requests between the enablers and the Sync API.
  
  
• 1 Engine : 1.5 CDR Enabler - To maintain a consistent workload on the engine queue for processing, we follow the guideline of making requests to process 1.5 files for every 1 engine available. This approach ensures a steady flow of work for the engine and helps optimize its utilization.

Once the appropriate number of replica counts has been determine - based on desired concurrency, calculate the necessary resources and adjust the auto-scale or manual value accordingly. This ensures that the system may handle the increased workload and efficiently allocate resources to meet the desired concurrency level for file processing in Constellations.