gRPC: creating Blocking or Future Stubs

Question

Lately, I 'm studying to build the gRPC client-server interaction.

I wrote a gRPC service :

service SearchService {

    rpc Find (SearchReq) returns (SearchRes);

}

and then I should invoke it on the client side using stub(another java-application).

whats the difference between this two types of stubs??

SearchServiceGrpc.SearchServiceFutureStub futureStub = SearchServiceGrpc.newFutureStub(channel);

SearchServiceGrpc.SearchServiceBlockingStub blockingStub = SearchServiceGrpc.newBlockingStub(channel);

blockingStub.find(request); - for blocking

futureStub.find(request).get(); - for future

I understand how works Futures in java, but I don't understand what happens inside gRPC and what type of stubs invocation is more productive

I find no information on google about it.

Thanks!

score 5 · Accepted Answer · answered Jan 25 '18 at 17:00

5

The various stubs are implemented using the same async API (ClientCall). Your choice doesn't matter for the grpc internals.

But the APIs have the normal expected limitations. For example: if you use blocking, that thread cannot be used until the RPC completes.

answered Jan 25 '18 at 17:00

Eric Anderson

24,057
5
55
76

it turns out that a future and blocking stub does not differ in performance except for a call in the client? where I can read the full docks? – lazarevsky Jan 26 '18 at 22:16
Future vs blocking does not impact the protocol used to talk to the server. – Eric Anderson Jan 29 '18 at 15:20
Thanks, now I understand! :D – lazarevsky Jul 30 '19 at 12:22

gRPC: creating Blocking or Future Stubs

1 Answers1