Rate limiting on the whole Apigee platform (on-premise with 2 DCs)

[christof1]

Hello,

In a on-premise deployment platform with 2 datacenters (and a load balancer fairly distributes request over the 2 DCs), how to implement this type of rate limiting:

- Define rates beyond which requests are blocked

- the rates could be defined by proxy, by app (which consumes my Apigee proxys), perhaps others...

- the main goal is to protect backends and also Apigee platform from unusually high traffic

- the rates must be considered on the whole apigee platform (shared by all Message Processors on the 2 data centers)

That being said, and regarding apigee capabilities, it is the spike arrest functionality which is asked, and not quota (example: limit to 300 req/s on a proxy). 

But the problem is to share these rates  between all MPs as the SpikeArrest policy stores the last time per message processor...

What do you think on that? What solutions could I look at?

Thank you very much for your help,

Regards

[dchiesa1]

That's a limitation in the SpikeArrest policy in OPDK. 

We've updated the SpikeArrest policy in Apigee X and hybrid, it now has better behavior in these kinds of cases (300 rps etc). 

I don't have a solution for you other than... build your own high-throughput rate limiter that works at 300 rps and in a distributed (multi DC) fashion. Not an easy task, I know.  

Sorry , I don't have better news.

[christof1]

Hello,

Thank you for this answer.

Do you know if this subject is planned in the OPDK roadmap?

Regards

[dchiesa1]

I do not believe we have published a roadmap for OPDK that includes this rate limiter.