On premise Setup: Any estimation template or method available for sizing the hardware requirement

[Chettiappan]

For APIGEE on premise setup, what is the best way to arrive a decision on sizing based on average TPS, peak TPS, payload size? In the documentation, I find details about the minimum hardware requirement.

[Denis_KALITVI]

You can try and see these (I guess you already been there): https://docs.apigee.com/private-cloud/v4.50.00/installation-requirements

About exact sizing, there's not public tool avail. AFAIK. Sizing depends on so many factors, like latency, protocol(SOAP\REST), tls\plain\mtls, size of payload, number of network jumps etc. I learned to build sizing based on my past customer projects +- estimation. I suggest you maybe contact support with detailed numbers to see if they can help. 

[sillan_dt]

I have designed multiple apigee architectures for many organizations. 

- the first thing you need to find is current TPS

- How much is expected in a spike

- what is the expected TPS in next five years

- how many regions/ datacenters required.

- Disaster recovery plan

- Once this is designed, you need to do proper load test. 

As apigee is having the customised Architecture option, you can add more RMPs and cassandras later.

I would suggest to have Postgres of minimum 3TB.

Cassandras of 1TB, 16 CPU

RMPS of 500GB, 16 CPU

Just calculate like 100TPS one RMP for better performance and stability.  you can have 3 cassandras for each 8-10 RMPs.

Payload default is 10MB and I don't think you need to allow more than that. For organisational security you can restrict to 6MB.