I assume you mean some kind of pipelining so that if multiple requests that have the same signature come through then only one gets executed and the response is shared amongst all the requesting clients without them ever touching the underlying application?
This could be an interesting use case indeed, but what I need is more like what's described in this paper
http://www.infoq.com/minibooks/emag-microservices
Page 8.
So the usecase is - you have a client that has to call say 8 microservices, but doesn't want to invoke them one by one or in pararell due to nature of HTTP protocol. Rather it wants to execute just 1 request with 8 calls (batch them) and get returned single response that encapsulates 8 responses.
Got it, yes that would indeed be very useful for heavy-use APIs, and shouldn't be too much of a problem to implement.
However it would essentially add a default batch endpoint to an API which isn't defined by the owner. Also, certain things like maximum batched requests would need to be implemented. Also, how would this affect quotas and RPM? Many questions, but well worth investigating, will add to roadmap.
I assume you mean some kind of pipelining so that if multiple requests that have the same signature come through then only one gets executed and the response is shared amongst all the requesting clients without them ever touching the underlying application?