yunfengzhou-hub commented on code in PR #97: URL: https://github.com/apache/flink-ml/pull/97#discussion_r874532378
########## flink-ml-iteration/src/main/java/org/apache/flink/iteration/datacache/nonkeyed/MemorySegmentWriter.java: ########## @@ -0,0 +1,86 @@ +/* + * 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.flink.iteration.datacache.nonkeyed; + +import org.apache.flink.annotation.Internal; +import org.apache.flink.core.fs.Path; +import org.apache.flink.runtime.memory.MemoryAllocationException; +import org.apache.flink.runtime.memory.MemoryManager; +import org.apache.flink.runtime.memory.MemoryReservationException; +import org.apache.flink.util.Preconditions; + +import org.openjdk.jol.info.GraphLayout; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.Optional; + +/** A class that writes cache data to memory segments. */ +@Internal +public class MemorySegmentWriter<T> implements SegmentWriter<T> { + private final Segment segment; + + private final MemoryManager memoryManager; + + public MemorySegmentWriter(Path path, MemoryManager memoryManager) + throws MemoryAllocationException { + this(path, memoryManager, 0L); + } + + public MemorySegmentWriter(Path path, MemoryManager memoryManager, long expectedSize) + throws MemoryAllocationException { + Preconditions.checkNotNull(memoryManager); + this.segment = new Segment(); + this.segment.path = path; + this.segment.cache = new ArrayList<>(); + this.segment.inMemorySize = 0L; + this.memoryManager = memoryManager; + } + + @Override + public boolean addRecord(T record) throws IOException { + if (!MemoryUtils.isMemoryEnoughForCache(memoryManager)) { + return false; + } + + long recordSize = GraphLayout.parseInstance(record).totalSize(); Review Comment: Spark uses its own classes, like `SizeTracker` and `SizeEstimator`, to calculate the size of each object. If we follow Spark's practice, we would also need to add such infra codes, which could add certain workload to this PR. `GraphLayout` uses `java.lang.instrument.Instrumentation`, which is the default solution for Java programs to get object size, so I think its overhead should be acceptable. The reason why I did not use `Instrumentation` is because that would require users add extra java options while executing flink ml programs, for example, ```shell java -javaagent:target/flink-ml-agent-2.1-SNAPSHOT.jar flink-ml-uber.jar <className> ``` The `-javaagent` option is additionally required. This more complex cmd requirement would decrease user experience while not bringing much performance improvement. -- 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]
