szetszwo commented on PR #6613: URL: https://github.com/apache/ozone/pull/6613#issuecomment-3354415164
@chungen0126 , thanks a lot for working on this! Looked at the existing code vs this change and had the following observations: --- Current with readChunk code: - Server side slowness: ⚠️ In the ChunkUtils code, we open and close the local file for each chunk. It will be slow for large files. - Client side slowness: ⚠️ readChunk(..) -> XceiverClientSpi.sendCommand(..) -> XceiverClientGrpc.sendCommandAsync(..) but XceiverClientSpi.sendCommand(..) will wait for the future for each chunk (i.e. it is NOT aysnc) ⚠️ In XceiverClientGrpc.sendCommandAsync(..) , we create a new gRPC stream, read a chunk and then close it. --- This pull request: - Server side: ⚠️ Still using ChunkUtils and open/close the local file for each chunk. - Client side: ✅ Using gRPC async stream (i.e. it is real async) ½✅ sendCommandReadBlock(..) only creates a gRPC stream per read(..) call (not a single stream per block) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
