Hi Everyone, So this is an opportunity to work as a community to explore solutions to improve gRPC-Java <https://github.com/grpc/grpc-java> when one has large payloads.
You might have seen the following post by Carl of a slowdown with large payloads using gRPC-Java <https://github.com/grpc/grpc-java>: https://github.com/grpc/grpc-java/issues/2151 You'll notice there is a lot of copying that is happening, which has a potential for being streamlined. In theory the passing-by-reference with the functional programming approach of updating the data is a natural direction, but currently different components of the program have specific object types they understand and expect for processing. There are some nice discussions happening around multi-buffer allocation, consolidated buffers before a write, zero-copy for the data portion with WriteCombiningHandler <https://github.com/buchgr/grpc-java/blob/08c67f80d6ff33e350e02891a5fc9f067cf192bd/netty/src/main/java/io/grpc/netty/WriteCombiningHandler.java> performing the serialization: https://github.com/grpc/grpc-java/pull/2092 https://github.com/grpc/grpc-java/pull/1996 https://github.com/grpc/grpc-java/issues/2139 This is a fun problem, which I imagine others might encounter when using gRPC-Java <https://github.com/grpc/grpc-java>. So let's explore ways to optimize this for large payloads, and discuss tweaks to gRPC-Java <https://github.com/grpc/grpc-java> to make this possible. Thanks, Paul -- You received this message because you are subscribed to the Google Groups "grpc.io" group. To unsubscribe from this group and stop receiving emails from it, send an email to grpc-io+unsubscr...@googlegroups.com. To post to this group, send email to grpc-io@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/grpc-io/49ac302b-0271-4fcb-99fd-0ca2779b49ad%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
