1. 简介
将Java对象进行缓存或者在网络间传输时，Kryo用于将Java对象序列化成为字节数组，或将字节数组反序列化成Java对象。
2. 优点
序列化后的结果小；功能完善；支持用户自定序列化器，注册类的序列化器非常方便。
3. 示例与对比
定义类：学生
public class Student implements Serializable {
private String name;
private int age;
private boolean isMale;
public Student(){}
public Student(String name, int age, boolean isMale) {
this.name = name;
this.age = age;
this.isMale = isMale;
}
}
方法一：用JDK自带的序列化功能
Student student = new Student("张三", 18, true);
//serialize
ByteArrayOutputStream bos = new ByteArrayOutputStream();
ObjectOutputStream out = new ObjectOutputStream(bos);
out.writeObject(student);
out.close();
byte[] bytes = bos.toByteArray();
//deserialize
ByteArrayInputStream bis = new ByteArrayInputStream(bytes);
ObjectInputStream in = new ObjectInputStream(bis);
Student newStudent = (Student) in.readObject();
in.close();
assert newStudent.equals(student);
方法二：用Kryo原生的序列化反序列化功能
Student student = new Student("张三", 18, true);
Kryo kryo = new Kryo();
//serialize
ByteArrayOutputStream bos = new ByteArrayOutputStream();
Output output = new Output(bos);
kryo.writeObject(output, student);
output.flush();
byte[] bytes = bos.toByteArray();
//deserialize
Input input = new Input(bytes);
Student newStudent = kryo.readObject(input, Student.class);
assert student.equals(newStudent);
方法三：自定义Student的序列化器，并注册到Kryo中
public class StudentSerializer extends Serializer<Student> {
@Override
public void write(Kryo kryo, Output output, Student object) {
output.writeString(object.getName());
output.writeVarInt(object.getAge(), true);
output.writeBoolean(object.isMale());
}
@Override
public Student read(Kryo kryo, Input input, Class<Student> type) {
String name = input.readString();
int age = input.readVarInt(true);
boolean isMale = input.readBoolean();
return new Student(name, age, isMale);
}
}
向Kryo注册序列化器
Kryo kryo = new Kryo();
kryo.register(Student.class, new StudentSerializer());
序列化和反序列化与方法二一致。
4. 序列化效果对比
| 方法 | 序列化字节长度 | 序列化耗时(ms) | 反序列化耗时（ms） |
|---|---|---|---|
| JDK序列化 | 135 | 19 | 1 |
| Kryo序列化 | 11 | 18 | 14 |
| Kryo注册序列化器 | 10 | 0 | 0 |
5. Kryo核心类图与基本用法
Kryo核心类图
Kryo对Class的序列化：
kryo.writeClass(Output output, Class type)
kryo.readClass(Input input)
//原理：初次序列化类名，后续序列化Registeration的ID
Kryo对Object的序列化：
kryo.writeObject(Output output, Object object)
kryo.readObject<T>(Input input, Class<T> type)
//原理：先根据Class信息获取Registeration，并获取对应的Serializer
// 然后调用Serializer的read和write方法
kryo.writeObjectOrNull(Output output, Object object，Class type)
kryo.readObjectOrNull<T>(Input input, Class<T> type)
//原理：写入每个对象的序列化结果之前，先往buffer中写入一个boolean用来表示当前对象是否为null，序列化结果会更大
kryo.writeClassAndObject(Output output, Object object)
kryo.readClassAndObject<T>(Input input)
//写入对象和对象的类信息，用于在反序列化时不知道对象Class的情况
6. Kryo对多引用对象的优化：kryo.setReferences(true)
优点：当被序列化对象（grade）中存在对另外一个对象(student)的多次引用时，只序列化引用对象一次，后面只是序列化该对象的ID
缺点：需要为每个Object绑定自增ID，并维护其Mapping。
//初始化参数和注册序列化器
Kryo kryo = new Kryo();
kryo.setReferences(true);
kryo.register(Student.class, new StudentSerializer());
kryo.register(Grade.class, new GradeSerializer());
//创建对象
Student student = new Student("张三", 18, true);
List<Student> students = new ArrayList<Student>(10);
int size = 10;
while(size -- >0){
students.add(student);
}
Grade grade = new Grade("Grade1",students);
//序列化
...
kryo.writeObject(output, grade);
...
//反序列化
...
Grade newGrade = kryo.readObject(input, Grade.class);
...
Grade类
public class Grade {
private String gradeName;
private List<Student> students;
}
reference=false
reference=true
