Eikasia30

@ssz1997 For your visibility here.

A tentative run of this benchmark shows that `ReadResponseMarshaller` actually does worse on the `drainTo` method than the baseline implementation. It does improved a lot over the `read` method on...

Above is the benchmark for unmarshalling performance. Counting all the necessary steps of `ReadResponseMarshaller`, its throughput advantage over the baseline is really negligible. Either there's serious flaw in my benchmark...

I've run some experimental grpc read benchmarks using https://github.com/TachyonNexus/spark-dfsio on my macbook. The result suggests that there's no significant performance gain in using zero-copy either for worker or for client....

Re-runed with `ByteArrayOutputStream` as final consumer of serialized stream. It can be seen that `marshalZeroCopy` is still not better than `marshalBaselineDrain`

After a second look into the code, I realized I've made a big mistake in this benchmark so the performance gain cannot be seen: `DataMessageMarshaller` is designed to work with...

I think both "reuse" and "release" are feasible ideas, and I'm happy to work on a patch for this issue: - "Release" is relatively easy to implement. There is absolutely...

In order for `DefaultBlockWorkerClient` to inherit from `AbstractClient`, three issues have to be addressed: 1. `DefaultBlockWorkerClient` might not work with an `InetSocketAddress` as it might use a domain socket instead....

@beinan @dbw9580 Do you mind taking a look at this refactor? Thanks a lot!

@dbw9580 Cleaned up a bit and addressed your concerns. Would you like to take another look? Thanks!