EarthChen commented on code in PR #15983: URL: https://github.com/apache/dubbo/pull/15983#discussion_r2675710790
########## dubbo-rpc/dubbo-rpc-triple/src/main/java/org/apache/dubbo/rpc/protocol/tri/transport/TripleHttp2RemoteFlowController.java: ########## @@ -0,0 +1,73 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.dubbo.rpc.protocol.tri.transport; + +import io.netty.handler.codec.http2.DefaultHttp2RemoteFlowController; +import io.netty.handler.codec.http2.Http2Connection; +import io.netty.handler.codec.http2.Http2RemoteFlowController; +import io.netty.handler.codec.http2.StreamByteDistributor; +import io.netty.handler.codec.http2.WeightedFairQueueByteDistributor; + +/** + * Triple-specific implementation of {@link Http2RemoteFlowController}. + * + * <p>This class extends the {@link DefaultHttp2RemoteFlowController} to provide + * flow control management for Triple (Dubbo's gRPC-compatible protocol) connections. + * It coordinates the distribution of outbound flow-controlled bytes across all + * active streams within a connection. + * + * <p>The controller utilizes a {@link WeightedFairQueueByteDistributor} to ensure + * that bandwidth is allocated based on stream weights and priorities while + * maintaining fairness to prevent stream starvation. + * + * @see DefaultHttp2RemoteFlowController + * @see WeightedFairQueueByteDistributor + */ +public class TripleHttp2RemoteFlowController extends DefaultHttp2RemoteFlowController { + + /** + * Constructs a new TripleHttp2RemoteFlowController. + * + * @param connection the {@link Http2Connection} to be managed. + * @param streamByteDistributor the distributor responsible for determining how + * available bytes are allocated among streams. + */ + public TripleHttp2RemoteFlowController(Http2Connection connection, StreamByteDistributor streamByteDistributor) { + super(connection, streamByteDistributor); + } + + /** + * Factory method to create a pre-configured flow controller optimized for Triple performance. + * + * <p>Configuration details: + * <ul> + * <li>Uses {@link WeightedFairQueueByteDistributor} for weighted-fair resource allocation.</li> + * <li>Sets {@code allocationQuantum} to 16KB. This setting reduces scheduling overhead + * by ensuring each stream is allocated a meaningful chunk of data before switching + * contexts, which significantly improves throughput in high-load scenarios.</li> + * </ul> + * + * @param connection the {@link Http2Connection} for which the controller will be created. + * @return a fully initialized {@link Http2RemoteFlowController} instance. + */ + public static Http2RemoteFlowController newController(Http2Connection connection) { + WeightedFairQueueByteDistributor dist = new WeightedFairQueueByteDistributor(connection); Review Comment: No! Example: Suppose you have a connection performing two operations simultaneously: 1. **Stream A:** Downloading a 1GB log file (Weight: 16). 2. **Stream B:** A very short control command "STOP_SERVICE" (Weight: 16). * **With `Uniform`:** It sends 16KB of logs, then 16KB of the command, cycling back and forth. If the network is congested, the critical control command might have to wait in line for a significant amount of time behind the massive log data. * **With `WFQ`:** It can be configured to give Stream B a higher weight. Even if the weights are equal, the WFQ scheduling algorithm manages the "virtual clock" more precisely when handling concurrent streams, ensuring that the control command can be interleaved and dispatched through the data stream at the fastest possible speed. -- 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]
