Recommendation for number of RMP nodes

joshdij
Participant I

In the Apigee Private Cloud, what does Apigee recommend the number of RMP nodes to be provided (through a load balancer)?

For example, "for every 5000 TPS, add another RMP node" or "for every 10,000,000 requests in 5 minutes, add another RMP node", etc.

Any insight or documentation would be greatly appreciated!

0 2 363
2 REPLIES 2

joshdij
Participant I

For this scenario, please assume that the Router Message Processor follows the guidelines at Installation requirements

During ops training, a common number was in the region of 1000TPS per message processor hosting "medium complexity proxies".

In the real world, with factors such as number of orgs/envs/proxies deployed and your segregation of hardware resources, this number can vary hugely. This is further complicated by other factors such as response times from back end systems etc.

Obviously, pass-through proxies to fast backends will result in much higher throughputs than proxies that use a large number of policies, payload transformations etc.

The message processors and routers can be tuned extensively to your particular workloads.

The recommendation, as always, is to run performance tests on your system using the same proxies in the same ratio as you'd expect in production.

https://community.apigee.com/articles/36324/apigee-private-cloud-performance-optimization.html