[jira] [Commented] (FLINK-1320) Add an off-heap variant of the managed memory

["ASF GitHub Bot (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-1320?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14301746#comment-14301746 ] 

ASF GitHub Bot commented on FLINK-1320:
---------------------------------------

Github user hsaputra commented on a diff in the pull request:

    https://github.com/apache/flink/pull/290#discussion_r23950666
  
    --- Diff: flink-core/src/main/java/org/apache/flink/core/memory/DirectMemorySegment.java ---
    @@ -0,0 +1,595 @@
    +/*
    + * 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.core.memory;
    +
    +import java.io.DataInput;
    +import java.io.DataOutput;
    +import java.io.IOException;
    +import java.lang.reflect.Field;
    +import java.nio.BufferOverflowException;
    +import java.nio.BufferUnderflowException;
    +import java.nio.ByteBuffer;
    +
    +/**
    + * This class uses in parts code from Java's direct byte buffer API.
    + * 
    + * The use in this class two crucial additions:
    + *  - It uses collapsed checks for range check and memory segment disposal.
    + *  - It offers absolute positioning methods for byte array put/get methods, to guarantee thread safe use.
    + *  
    + * In addition, the code that uses this class should make sure that only one implementation class is ever loaded -
    + * Either the {@link HeapMemorySegment}, or this DirectMemorySegment. That way, all the abstract methods in the
    + * MemorySegment base class have only one loaded actual implementation. This is easy for the JIT to recognize through
    + * class hierarchy analysis, or by identifying that the invocations are monomorphic (all go to the same concrete
    + * method implementation). Under this precondition, the JIT can perfectly inline methods.
    + * 
    + * This is harder to do and control with byte buffers, where different code paths use different versions of the class
    + * (heap, direct, mapped) and thus virtual method invocations are polymorphic and are not as easily inlined.
    + */
    +public class DirectMemorySegment extends MemorySegment {
    +	
    +	/** The direct byte buffer that allocated the memory */
    +	protected final ByteBuffer buffer;
    +	
    +	/** The address to the off-heap data */
    +	private long address;
    +	
    +	/** The address one byte after the last addressable byte.
    +	 *  This is address + size while the segment is not disposed */
    +	private final long addressLimit;
    +	
    +	/** The size in bytes of the memory segment */
    +	private final int size;
    +	
    +	// -------------------------------------------------------------------------
    +	//                             Constructors
    +	// -------------------------------------------------------------------------
    +
    +	public DirectMemorySegment(int size) {
    +		this(ByteBuffer.allocateDirect(size));
    +	}
    +
    +	public DirectMemorySegment(ByteBuffer buffer) {
    +		if (buffer == null || !buffer.isDirect()) {
    +			throw new IllegalArgumentException();
    --- End diff --
    
    Put message to the exception to help diagnose problem.


> Add an off-heap variant of the managed memory
> ---------------------------------------------
>
>                 Key: FLINK-1320
>                 URL: https://issues.apache.org/jira/browse/FLINK-1320
>             Project: Flink
>          Issue Type: Improvement
>          Components: Local Runtime
>            Reporter: Stephan Ewen
>            Assignee: Max Michels
>            Priority: Minor
>
> For (nearly) all memory that Flink accumulates (in the form of sort buffers, hash tables, caching), we use a special way of representing data serialized across a set of memory pages. The big work lies in the way the algorithms are implemented to operate on pages, rather than on objects.
> The core class for the memory is the {{MemorySegment}}, which has all methods to set and get primitives values efficiently. It is a somewhat simpler (and faster) variant of a HeapByteBuffer.
> As such, it should be straightforward to create a version where the memory segment is not backed by a heap byte[], but by memory allocated outside the JVM, in a similar way as the NIO DirectByteBuffers, or the Netty direct buffers do it.
> This may have multiple advantages:
>   - We reduce the size of the JVM heap (garbage collected) and the number and size of long living alive objects. For large JVM sizes, this may improve performance quite a bit. Utilmately, we would in many cases reduce JVM size to 1/3 to 1/2 and keep the remaining memory outside the JVM.
>   - We save copies when we move memory pages to disk (spilling) or through the network (shuffling / broadcasting / forward piping)
> The changes required to implement this are
>   - Add a {{UnmanagedMemorySegment}} that only stores the memory adress as a long, and the segment size. It is initialized from a DirectByteBuffer.
>   - Allow the MemoryManager to allocate these MemorySegments, instead of the current ones.
>   - Make sure that the startup script pick up the mode and configure the heap size and the max direct memory properly.
> Since the MemorySegment is probably the most performance critical class in Flink, we must take care that we do this right. The following are critical considerations:
>   - If we want both solutions (heap and off-heap) to exist side-by-side (configurable), we must make the base MemorySegment abstract and implement two versions (heap and off-heap).
>   - To get the best performance, we need to make sure that only one class gets loaded (or at least ever used), to ensure optimal JIT de-virtualization and inlining.
>   - We should carefully measure the performance of both variants. From previous micro benchmarks, I remember that individual byte accesses in DirectByteBuffers (off-heap) were slightly slower than on-heap, any larger accesses were equally good or slightly better.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)