并发主题

各种Java序列化性能比较

这里比较Java对象序列化 XML JSON Kryo POF等序列化性能比较。

很多人以为JDK的Java序列化肯定是将Java对象转换成二进制序列化最快的方式,JDK7出来以后,我们发现实际上每次新的JDK比旧版本快。

我们通常以为将Java对象序列化成二进制比序列化成XML或Json更快,其实是错误的,如果你关心性能,建议避免Java序列化。

Java序列化有很多的要求,最主要的一个是包含能够序列化任何东西(或至少任何实现Serializable接口)。这样才能进入其他JVM之中,这很重要,所以有时性能不是主要的要求,标准的格式才最重要。

我们经常看到CPU花费很多时间内进行Java序列化,下面我们研究一下,假设一定Order,虽然只有几个字节,但是序列化以后不是几十个字节,而是600多个字节:

Ordr代码:

public class Order implements Serializable {
    private long id;
    private String description;
    private BigDecimal totalCost = BigDecimal.valueOf(0);
    private List orderLines = new ArrayList();
    private Customer customer;

...

}

序列化输出:

----sr--model.Order----h#-----J--idL--customert--Lmodel/Customer;L--descriptiont--Ljava/lang/String;L--orderLinest--Ljava/util/List;L--totalCostt--Ljava/math/BigDecimal;xp--------ppsr--java.util.ArrayListx-----a----I--sizexp----w-----sr--model.OrderLine--&-1-S----I--lineNumberL--costq-~--L--descriptionq-~--L--ordert--Lmodel/Order;xp----sr--java.math.BigDecimalT--W--(O---I--scaleL--intValt--Ljava/math/BigInteger;xr--java.lang.Number-----------xp----sr--java.math.BigInteger-----;-----I--bitCountI--bitLengthI--firstNonzeroByteNumI--lowestSetBitI--signum[--magnitudet--[Bxq-~----------------------ur--[B------T----xp----xxpq-~--xq-~--

正如你可能已经注意到,Java序列化写入不仅是完整的类名,也包含整个类的定义,包含所有被引用的类。类定义可以是相当大的,也许构成了性能和效率的问题,当然这是编写一个单一的对象。如果您正在编写了大量相同的类的对象,这时类定义的开销通常不是一个大问题。另一件事情是,如果你的对象有一类的引用(如元数据对象),那么Java序列化将写入整个类的定义,不只是类的名称,因此,使用Java序列化写出元数据(meta-data)是非常昂贵的。

Externalizable

通过实现Externalizable接口,这是可能优化Java序列化的。实现此接口,避免写出整个类定义,只是类名被写入。它需要你实施readExternal和writeExternal方法方法的,所以需要做一些工作,但相比仅仅是实现Serializable更快,更高效。

Externalizable对小数目对象有效的多。但是对大量对象,或者重复对象,则效率低。

public class Order implements Externalizable {
    private long id;
    private String description;
    private BigDecimal totalCost = BigDecimal.valueOf(0);
    private List orderLines = new ArrayList();
    private Customer customer;

    public Order() {
    }

    public void readExternal(ObjectInput stream) throws IOException, ClassNotFoundException {
        this.id = stream.readLong();
        this.description = (String)stream.readObject();
        this.totalCost = (BigDecimal)stream.readObject();
        this.customer = (Customer)stream.readObject();
        this.orderLines = (List)stream.readObject();
    }

    public void writeExternal(ObjectOutput stream) throws IOException {
        stream.writeLong(this.id);
        stream.writeObject(this.description);
        stream.writeObject(this.totalCost);
        stream.writeObject(this.customer);
        stream.writeObject(this.orderLines);
    }
}

序列化输出:

----sr--model.Order---*3--^---xpw---------psr--java.math.BigDecimalT--W--(O---I--scaleL--intValt--Ljava/math/BigInteger;xr--java.lang.Number-----------xp----sr--java.math.BigInteger-----;-----I--bitCountI--bitLengthI--firstNonzeroByteNumI--lowestSetBitI--signum[--magnitudet--[Bxq-~----------------------ur--[B------T----xp----xxpsr--java.util.ArrayListx-----a----I--sizexp----w-----sr--model.OrderLine-!!|---S---xpw-----pq-~--q-~--xxx

EclipseLink MOXy - XML 和 JSON

序列化成XML或JSON可以允许其他语言访问,可以实现REST服务等。缺点是文本格式的效率比优化的二进制格式低一些,使用JAXB,你需要使用JAXB注释类,或提供一个XML配置文件。使用@XmlIDREF处理循环。

@XmlRootElement
public class Order {
    @XmlID
    @XmlAttribute
    private long id;
    @XmlAttribute
    private String description;
    @XmlAttribute
    private BigDecimal totalCost = BigDecimal.valueOf(0);
    private List orderLines = new ArrayList();
    private Customer customer;
}

public class OrderLine {
    @XmlIDREF
    private Order order;
    @XmlAttribute
    private int lineNumber;
    @XmlAttribute
    private String description;
    @XmlAttribute
    private BigDecimal cost = BigDecimal.valueOf(0);
}

XML输出:

<order id="0" totalCost="0">
<orderLines lineNumber="1" cost="0">
<order>0</order
></orderLines
></order>

JSOn输出:

{"order":{"id":0,"totalCost":0,"orderLines":[{"lineNumber":1,"cost":0,"order":0}]}}

Kryo

Kryo 是一种快速,高效的序列化的Java框架。 KRYO是新的BSD许可下一个开源项目提供。这是一个很小的项目,只有3名成员,它首先在2009年出品。

工作原理类似于Java序列化KRYO,尊重瞬态字段,但不要求一类是可序列化的。KRYO有一定的局限性,比如需要有一个默认的构造函数的类,在序列化将java.sql.Time java.sql.Date java.sql.Timestamp类会遇到一些问题。

order序列化结果:

------java-util-ArrayLis-----model-OrderLin----java-math-BigDecima---------model-Orde-----

Oracle Coherence POF

 Oracle Coherence 产品提供其自己优化的二进制格式,称为POF (可移植对象格式) 。 Oracle Coherence的是一个内存中的数据网格解决方案(分布式缓存) 。是一个商业产品,并需要许可证。

POF提供了一个序列化框架,并可以独立使用。 POF要求类实现一个PortableObject接口和读/写方法。您还可以实现一个单独的序列化类,或使用最新版本的序列化的注解。 POF要求每个类都被提前分配一个固定ID,所以你需要通过某种方式确定这个ID 。 POF格式是二进制格式,非常紧凑,高效,快速的,但确实需要你付出一些工作。

POF的总字节数为一个单一的订单/订单行对象为32个字节, 1593字节100 OrderLines的。我不会放弃的结果, POF是一个商业许可产品的一部分,但是是非常快的。

public class Order implements PortableObject {
    private long id;
    private String description;
    private BigDecimal totalCost = BigDecimal.valueOf(0);
    private List orderLines = new ArrayList();
    private Customer customer;

    public Order() {
    }

    public void readExternal(PofReader in) throws IOException {
        this.id = in.readLong(0);
        this.description = in.readString(1);
        this.totalCost = in.readBigDecimal(2);
        this.customer = (Customer)in.readObject(3);
        this.orderLines = (List)in.readCollection(4, new ArrayList());
    }

    public void writeExternal(PofWriter out) throws IOException {
        out.writeLong(0, this.id);
        out.writeString(1, this.description);
        out.writeBigDecimal(2, this.totalCost);
        out.writeObject(3, this.customer);
        out.writeCollection(4, this.orderLines);
    }
}

序列化结果:

-----B--G---d-U------A--G-------

 

性能比较

一个订单包含一个Oderline

Serializer Size (bytes) Serialize (operations/second) Deserialize (operations/second) % Difference (from Java serialize) % Difference (deserialize)
Java Serializable 636 128,634 19,180 0% 0%
Java Externalizable 435 160,549 26,678 24% 39%
EclipseLink MOXy XML 101 348,056 47,334 170% 146%
Kryo 90 359,368 346,984 179% 1709%

一个订单100个oderlines:


Serializer Size (bytes) Serialize (operations/second) Deserialize (operations/second) % Difference (from Java serialize) % Difference (deserialize)
Java Serializable 2,715 16,470 10,215 0% 0%
Java Externalizable 2,811 16,206 11,483 -1% 12%
EclipseLink MOXy XML 6,628 7,304 2,731 -55% -73%
Kryo 1216 22,862 31,499 38% 208%


本教程代码下载

要获得象C那样的序列化性能,直接自己编写。

Serialization ByteBuffer Unsafe三者性能比较:

三者性能测试代码:

import sun.misc.Unsafe;
import java.io.ByteArrayInputStream;
import java.io.ByteArrayOutputStream;
import java.io.ObjectInputStream;
import java.io.ObjectOutputStream;
import java.io.Serializable;
import java.lang.reflect.Field;
import java.nio.ByteBuffer;
import java.util.Arrays;

public final class TestSerialisationPerf
{
    public static final int REPETITIONS = 1 * 1000 * 1000;

    private static ObjectToBeSerialised ITEM =
        new ObjectToBeSerialised(
            1010L, true, 777, 99,
            new double[]{0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0},
            new long[]{1, 2, 3, 4, 5, 6, 7, 8, 9, 10});

 

    public static void main(final String[] arg) throws Exception
    {
        for (final PerformanceTestCase testCase : testCases)
        {
            for (int i = 0; i < 5; i++)
            {
                testCase.performTest();

                System.out.format('%d %s\twrite=%,dns read=%,dns total=%,dns\n',
                                  i,
                                  testCase.getName(),
                                  testCase.getWriteTimeNanos(),
                                  testCase.getReadTimeNanos(),
                                  testCase.getWriteTimeNanos() +
                                  testCase.getReadTimeNanos());

                if (!ITEM.equals(testCase.getTestOutput()))
                {
                    throw new IllegalStateException('Objects do not match');
                }

                System.gc();
                Thread.sleep(3000);
            }
        }
    }

    private static final PerformanceTestCase[] testCases =
    {
        new PerformanceTestCase('Serialisation', REPETITIONS, ITEM)
        {
            ByteArrayOutputStream baos = new ByteArrayOutputStream();

            public void testWrite(ObjectToBeSerialised item) throws Exception
            {
                for (int i = 0; i < REPETITIONS; i++)
                {
                    baos.reset();

                    ObjectOutputStream oos = new ObjectOutputStream(baos);
                    oos.writeObject(item);
                    oos.close();
                }
            }

            public ObjectToBeSerialised testRead() throws Exception
            {
                ObjectToBeSerialised object = null;
                for (int i = 0; i < REPETITIONS; i++)
                {
                    ByteArrayInputStream bais =
                        new ByteArrayInputStream(baos.toByteArray());
                    ObjectInputStream ois = new ObjectInputStream(bais);
                    object = (ObjectToBeSerialised)ois.readObject();
                }

                return object;
            }
        },

        new PerformanceTestCase('ByteBuffer', REPETITIONS, ITEM)
        {
            ByteBuffer byteBuffer = ByteBuffer.allocate(1024);

            public void testWrite(ObjectToBeSerialised item) throws Exception
            {
                for (int i = 0; i < REPETITIONS; i++)
                {
                    byteBuffer.clear();
                    item.write(byteBuffer);
                }
            }

            public ObjectToBeSerialised testRead() throws Exception
            {
                ObjectToBeSerialised object = null;
                for (int i = 0; i < REPETITIONS; i++)
                {
                    byteBuffer.flip();
                    object = ObjectToBeSerialised.read(byteBuffer);
                }

                return object;
            }
        },

        new PerformanceTestCase('UnsafeMemory', REPETITIONS, ITEM)
        {
            UnsafeMemory buffer = new UnsafeMemory(new byte[1024]);

            public void testWrite(ObjectToBeSerialised item) throws Exception
            {
                for (int i = 0; i < REPETITIONS; i++)
                {
                    buffer.reset();
                    item.write(buffer);
                }
            }

            public ObjectToBeSerialised testRead() throws Exception
            {
                ObjectToBeSerialised object = null;
                for (int i = 0; i < REPETITIONS; i++)
                {
                    buffer.reset();
                    object = ObjectToBeSerialised.read(buffer);
                }

                return object;
            }
        },
    };
}

abstract class PerformanceTestCase
{
    private final String name;
    private final int repetitions;
    private final ObjectToBeSerialised testInput;
    private ObjectToBeSerialised testOutput;
    private long writeTimeNanos;
    private long readTimeNanos;

    public PerformanceTestCase(final String name, final int repetitions,
                               final ObjectToBeSerialised testInput)
    {
        this.name = name;
        this.repetitions = repetitions;
        this.testInput = testInput;
    }

    public String getName()
    {
        return name;
    }

    public ObjectToBeSerialised getTestOutput()
    {
        return testOutput;
    }

    public long getWriteTimeNanos()
    {
        return writeTimeNanos;
    }

    public long getReadTimeNanos()
    {
        return readTimeNanos;
    }

    public void performTest() throws Exception
    {
        final long startWriteNanos = System.nanoTime();
        testWrite(testInput);
        writeTimeNanos = (System.nanoTime() - startWriteNanos) / repetitions;

        final long startReadNanos = System.nanoTime();
        testOutput = testRead();
        readTimeNanos = (System.nanoTime() - startReadNanos) / repetitions;
    }

    public abstract void testWrite(ObjectToBeSerialised item) throws Exception;
    public abstract ObjectToBeSerialised testRead() throws Exception;
}

class ObjectToBeSerialised implements Serializable
{
    private static final long serialVersionUID = 10275539472837495L;

    private final long sourceId;
    private final boolean special;
    private final int orderCode;
    private final int priority;
    private final double[] prices;
    private final long[] quantities;

    public ObjectToBeSerialised(final long sourceId, final boolean special,
                                final int orderCode, final int priority,
                                final double[] prices, final long[] quantities)
    {
        this.sourceId = sourceId;
        this.special = special;
        this.orderCode = orderCode;
        this.priority = priority;
        this.prices = prices;
        this.quantities = quantities;
    }

    public void write(final ByteBuffer byteBuffer)
    {
        byteBuffer.putLong(sourceId);
        byteBuffer.put((byte)(special ? 1 : 0));
        byteBuffer.putInt(orderCode);
        byteBuffer.putInt(priority);

        byteBuffer.putInt(prices.length);
        for (final double price : prices)
        {
            byteBuffer.putDouble(price);
        }

        byteBuffer.putInt(quantities.length);
        for (final long quantity : quantities)
        {
            byteBuffer.putLong(quantity);
        }
    }

    public static ObjectToBeSerialised read(final ByteBuffer byteBuffer)
    {
        final long sourceId = byteBuffer.getLong();
        final boolean special = 0 != byteBuffer.get();
        final int orderCode = byteBuffer.getInt();
        final int priority = byteBuffer.getInt();

        final int pricesSize = byteBuffer.getInt();
        final double[] prices = new double[pricesSize];
        for (int i = 0; i < pricesSize; i++)
        {
            prices[i] = byteBuffer.getDouble();
        }

        final int quantitiesSize = byteBuffer.getInt();
        final long[] quantities = new long[quantitiesSize];
        for (int i = 0; i < quantitiesSize; i++)
        {
            quantities[i] = byteBuffer.getLong();
        }

        return new ObjectToBeSerialised(sourceId, special, orderCode,
                                        priority, prices, quantities);
    }

    public void write(final UnsafeMemory buffer)
    {
        buffer.putLong(sourceId);
        buffer.putBoolean(special);
        buffer.putInt(orderCode);
        buffer.putInt(priority);
        buffer.putDoubleArray(prices);
        buffer.putLongArray(quantities);
    }

    public static ObjectToBeSerialised read(final UnsafeMemory buffer)
    {
        final long sourceId = buffer.getLong();
        final boolean special = buffer.getBoolean();
        final int orderCode = buffer.getInt();
        final int priority = buffer.getInt();
        final double[] prices = buffer.getDoubleArray();
        final long[] quantities = buffer.getLongArray();

        return new ObjectToBeSerialised(sourceId, special, orderCode,
                                        priority, prices, quantities);
    }

    @Override
    public boolean equals(final Object o)
    {
        if (this == o)
        {
            return true;
        }
        if (o == null || getClass() != o.getClass())
        {
            return false;
        }

        final ObjectToBeSerialised that = (ObjectToBeSerialised)o;

        if (orderCode != that.orderCode)
        {
            return false;
        }
        if (priority != that.priority)
        {
            return false;
        }
        if (sourceId != that.sourceId)
        {
            return false;
        }
        if (special != that.special)
        {
            return false;
        }
        if (!Arrays.equals(prices, that.prices))
        {
            return false;
        }
        if (!Arrays.equals(quantities, that.quantities))
        {
            return false;
        }

        return true;
    }
}

class UnsafeMemory
{
    private static final Unsafe unsafe;
    static
    {
        try
        {
            Field field = Unsafe.class.getDeclaredField('theUnsafe');
            field.setAccessible(true);
            unsafe = (Unsafe)field.get(null);
        }
        catch (Exception e)
        {
            throw new RuntimeException(e);
        }
    }

    private static final long byteArrayOffset = unsafe.arrayBaseOffset(byte[].class);
    private static final long longArrayOffset = unsafe.arrayBaseOffset(long[].class);
    private static final long doubleArrayOffset = unsafe.arrayBaseOffset(double[].class);

    private static final int SIZE_OF_BOOLEAN = 1;
    private static final int SIZE_OF_INT = 4;
    private static final int SIZE_OF_LONG = 8;

    private int pos = 0;
    private final byte[] buffer;

    public UnsafeMemory(final byte[] buffer)
    {
        if (null == buffer)
        {
            throw new NullPointerException('buffer cannot be null');
        }

        this.buffer = buffer;
    }

    public void reset()
    {
        this.pos = 0;
    }

    public void putBoolean(final boolean value)
    {
        unsafe.putBoolean(buffer, byteArrayOffset + pos, value);
        pos += SIZE_OF_BOOLEAN;
    }

    public boolean getBoolean()
    {
        boolean value = unsafe.getBoolean(buffer, byteArrayOffset + pos);
        pos += SIZE_OF_BOOLEAN;

        return value;
    }

    public void putInt(final int value)
    {
        unsafe.putInt(buffer, byteArrayOffset + pos, value);
        pos += SIZE_OF_INT;
    }

    public int getInt()
    {
        int value = unsafe.getInt(buffer, byteArrayOffset + pos);
        pos += SIZE_OF_INT;

        return value;
    }

    public void putLong(final long value)
    {
        unsafe.putLong(buffer, byteArrayOffset + pos, value);
        pos += SIZE_OF_LONG;
    }

    public long getLong()
    {
        long value = unsafe.getLong(buffer, byteArrayOffset + pos);
        pos += SIZE_OF_LONG;

        return value;
    }

    public void putLongArray(final long[] values)
    {
        putInt(values.length);

        long bytesToCopy = values.length << 3;
        unsafe.copyMemory(values, longArrayOffset,
                          buffer, byteArrayOffset + pos,
                          bytesToCopy);
        pos += bytesToCopy;
    }

    public long[] getLongArray()
    {
        int arraySize = getInt();
        long[] values = new long[arraySize];

        long bytesToCopy = values.length << 3;
        unsafe.copyMemory(buffer, byteArrayOffset + pos,
                          values, longArrayOffset,
                          bytesToCopy);
        pos += bytesToCopy;

        return values;
    }

    public void putDoubleArray(final double[] values)
    {
        putInt(values.length);

        long bytesToCopy = values.length << 3;
        unsafe.copyMemory(values, doubleArrayOffset,
                          buffer, byteArrayOffset + pos,
                          bytesToCopy);
        pos += bytesToCopy;
    }

    public double[] getDoubleArray()
    {
        int arraySize = getInt();
        double[] values = new double[arraySize];

        long bytesToCopy = values.length << 3;
        unsafe.copyMemory(buffer, byteArrayOffset + pos,
                          values, doubleArrayOffset,
                          bytesToCopy);
        pos += bytesToCopy;

        return values;
    }
}

测试结果:

2.8GHz Nehalem - Java 1.7.0_04
==============================
0 Serialisation write=2,517ns read=11,570ns total=14,087ns
1 Serialisation write=2,198ns read=11,122ns total=13,320ns
2 Serialisation write=2,190ns read=11,011ns total=13,201ns
3 Serialisation write=2,221ns read=10,972ns total=13,193ns
4 Serialisation write=2,187ns read=10,817ns total=13,004ns
0 ByteBuffer write=264ns read=273ns total=537ns
1 ByteBuffer write=248ns read=243ns total=491ns
2 ByteBuffer write=262ns read=243ns total=505ns
3 ByteBuffer write=300ns read=240ns total=540ns
4 ByteBuffer write=247ns read=243ns total=490ns
0 UnsafeMemory write=99ns read=84ns total=183ns
1 UnsafeMemory write=53ns read=82ns total=135ns
2 UnsafeMemory write=63ns read=66ns total=129ns
3 UnsafeMemory write=46ns read=63ns total=109ns
4 UnsafeMemory write=48ns read=58ns total=106ns

2.4GHz Sandy Bridge - Java 1.7.0_04
===================================
0 Serialisation write=1,940ns read=9,006ns total=10,946ns
1 Serialisation write=1,674ns read=8,567ns total=10,241ns
2 Serialisation write=1,666ns read=8,680ns total=10,346ns
3 Serialisation write=1,666ns read=8,623ns total=10,289ns
4 Serialisation write=1,715ns read=8,586ns total=10,301ns
0 ByteBuffer write=199ns read=198ns total=397ns
1 ByteBuffer write=176ns read=178ns total=354ns
2 ByteBuffer write=174ns read=174ns total=348ns
3 ByteBuffer write=172ns read=183ns total=355ns
4 ByteBuffer write=174ns read=180ns total=354ns
0 UnsafeMemory write=38ns read=75ns total=113ns
1 UnsafeMemory write=26ns read=52ns total=78ns
2 UnsafeMemory write=26ns read=51ns total=77ns
3 UnsafeMemory write=25ns read=51ns total=76ns
4 UnsafeMemory write=27ns read=50ns total=77ns

很显然允许自己内存操作的 Unsafe性能是最快的。

 

java多线程

Java同步或锁

Java性能调优

 

 

猜你喜欢