Deployment Guidelines for Optimal CPU Usage
This topic provides deployment strategies for achieving optimal CPU usage across various service configurations. It includes recommendations for scenarios where services are deployed on a single instance and when the Enterprise Messaging Service (EMS) is hosted on a separate instance.
Deployment Configurations
All services on the same instance
When deploying all services (EMS, Config, Broker, Orchestrator, Aopd, PC, Catalog) on the same instance, follow these guidelines:
-
Orchestrator (
internalProcessorSize
): Allocate resources equivalent to 1x the number of processors. -
Aopd (
planGenerationReceiverCount
): Allocate resources equivalent to 1/4x the number of processors.
For optimal performance, configure the instance with 36 vCPUs and 60 GB RAM.
Example: Deploy one OMS instance and three OMS services on a single instance.
EMS on a different instance
For setups where EMS is hosted on a different instance from other services, use the following configuration:
-
Orchestrator (
internalProcessorSize
): Allocate resources equivalent to 2x the number of processors. -
AOPD (
planGenerationReceiverCount
): Allocate resources equivalent to 1/2x the number of processors.
For optimal performance, configure the instance with 16 vCPUs and 30 GB RAM.
Example: Deploy three OMS instances distributed across two instances.