Java 缓存技术

以下仅是对map对方式讨论。没有对点阵图阵讨论。
作缓存要做以下2点: 

1:清理及更新缓存时机的处理:
   . 虚拟机内存不足,清理缓存
   .. 缓存时间超时,或访问次数超出, 启动线程更新
2:类和方法的反射 (线程嵌套调用)
   reflect.invoke的使用。

 

JAVA缓存有两种: 
一、文件缓存,是指把数据存储在磁盘上,可以XML格式,也可以序列化文件DAT格式还是其它文件格式。 
二、内存缓存,也就是实现一个类中静态Map,对这个Map进行常规的增删查。

 

 

代码如下:内存中的缓存

package atest.hiyaCache;

/**
 *
 * @author hiyachen
 * @version $Revision$
 */
public class AddrDetail {

    public String latelyKeyword(String province, String city, String county){
        System.out.println("AddrDetail.latelyKeyword=" + province + city + county);
        return province + city + county;
    }

    public String buildCache(String latelyKeyword){
        System.out.println("AddrDetail.buildCache=" + latelyKeyword);
        return latelyKeyword;
    }
}

 

package atest.hiyaCache;

public class CacheData {
    private Object data;
    private long time;
    private int count;

    public CacheData() {

    }

    public CacheData(Object data, long time, int count) {
        this.data = data;
        this.time = time;
        this.count = count;
    }

    public CacheData(Object data) {
        this.data = data;
        this.time = System.currentTimeMillis();
        this.count = 1;
    }

    public void addCount() {
        count++;
    }

    public int getCount() {
        return count;
    }
    public void setCount(int count) {
        this.count = count;
    }
    public Object getData() {
        return data;
    }
    public void setData(Object data) {
        this.data = data;
    }
    public long getTime() {
        return time;
    }
    public void setTime(long time) {
        this.time = time;
    }
}

 

 

package atest.hiyaCache;

import java.lang.reflect.Method;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Hashtable;

import org.apache.commons.logging.Log;   //commons-loggings-1.1.1.jar
import org.apache.commons.logging.LogFactory;

public class CacheOperation {
    private static final Log log = LogFactory.getLog(CacheOperation.class);
    private static CacheOperation singleton = null;

    private Hashtable cacheMap;//存放缓存数据

    private ArrayList threadKeys;//处于线程更新中的key值列表

    public static CacheOperation getInstance() {
        if (singleton == null) {
            singleton = new CacheOperation();
        }
        return singleton;
    }

    private CacheOperation() {
        cacheMap = new Hashtable();
        threadKeys = new ArrayList();
    }

    /**
     * 添加数据缓存
     * 与方法getCacheData(String key, long intervalTime, int maxVisitCount)配合使用
     * @param key
     * @param data
     */
    public void addCacheData(String key, Object data) {
        addCacheData(key, data, true);
    }

    private void addCacheData(String key, Object data, boolean check) {
        if (Runtime.getRuntime().freeMemory() < 5L*1024L*1024L) {//虚拟机内存小于10兆,则清除缓存
            log.warn("WEB缓存:内存不足,开始清空缓存!");
            removeAllCacheData();
            return;
        } else if(check && cacheMap.containsKey(key)) {
            log.warn("WEB缓存:key值= " + key + " 在缓存中重复, 本次不缓存!");
            return;
        }
        cacheMap.put(key, new CacheData(data));
    }

    /**
     * 取得缓存中的数据
     * 与方法addCacheData(String key, Object data)配合使用
     * @param key
     * @param intervalTime 缓存的时间周期,小于等于0时不限制
     * @param maxVisitCount 访问累积次数,小于等于0时不限制
     * @return
     */
    public Object getCacheData(String key, long intervalTime, int maxVisitCount) {
        CacheData cacheData = (CacheData)cacheMap.get(key);
        if (cacheData == null) {
            return null;
        }
        if (intervalTime > 0 && (System.currentTimeMillis() - cacheData.getTime()) > intervalTime) {
            removeCacheData(key);
            return null;
        }
        if (maxVisitCount > 0 && (maxVisitCount - cacheData.getCount()) <= 0) {
            removeCacheData(key);
            return null;
        } else {
            cacheData.addCount();
        }
        return cacheData.getData();
    }

    /**
     * 当缓存中数据失效时,用不给定的方法线程更新数据
     * @param o 取得数据的对像(该方法是静态方法是不用实例,则传Class实列)
     * @param methodName 该对像中的方法
     * @param parameters 该方法的参数列表(参数列表中对像都要实现toString方法,若列表中某一参数为空则传它所属类的Class)
     * @param intervalTime 缓存的时间周期,小于等于0时不限制
     * @param maxVisitCount 访问累积次数,小于等于0时不限制
     * @return
     */
    public Object getCacheData(Object o, String methodName,Object[] parameters,
            long intervalTime, int maxVisitCount) {
        Class oc = o instanceof Class ? (Class)o : o.getClass();
        StringBuffer key = new StringBuffer(oc.getName());//生成缓存key值
        key.append("-").append(methodName);
        if (parameters != null) {
            for (int i = 0; i < parameters.length; i++) {
                if (parameters[i] instanceof Object[]) {
                    key.append("-").append(Arrays.toString((Object[])parameters[i]));
                } else {
                    key.append("-").append(parameters[i]);
                }
            }
        }

        CacheData cacheData = (CacheData)cacheMap.get(key.toString());
        if (cacheData == null) {//等待加载并返回
            Object returnValue = invoke(o, methodName, parameters, key.toString());
            return returnValue instanceof Class ? null : returnValue;
        }
        if (intervalTime > 0 && (System.currentTimeMillis() - cacheData.getTime()) > intervalTime) {
            daemonInvoke(o, methodName, parameters, key.toString());//缓存时间超时,启动线程更新数据
        } else if (maxVisitCount > 0 && (maxVisitCount - cacheData.getCount()) <= 0) {//访问次数超出,启动线程更新数据
            daemonInvoke(o, methodName, parameters, key.toString());
        } else {
            cacheData.addCount();
        }
        return cacheData.getData();
    }

    /**
     * 递归调用给定方法更新缓存中数据据
     * @param o
     * @param methodName
     * @param parameters
     * @param key
     * @return 若反射调用方法返回值为空则返回该值的类型
     */
    private Object invoke(Object o, String methodName,Object[] parameters, String key) {
        Object returnValue = null;
        try {
            Class[] pcs = null;
            if (parameters != null) {
                pcs = new Class[parameters.length];
                for (int i = 0; i < parameters.length; i++) {
                    if (parameters[i] instanceof MethodInfo) {
//参数类型是MethodInfo则调用该方法的返回值做这参数
                        MethodInfo pmi = (MethodInfo)parameters[i];
                        Object pre = invoke(pmi.getO(), pmi.getMethodName(), pmi.getParameters(), null);
                        parameters[i] = pre;
                    }
                    if (parameters[i] instanceof Class) {
                        pcs[i] = (Class)parameters[i];
                        parameters[i] = null;
                    } else {
                        pcs[i] = parameters[i].getClass();
                    }
                }
            }
            Class oc = o instanceof Class ? (Class)o : o.getClass();
            //Method m = oc.getDeclaredMethod(methodName, pcs);
            Method m = matchMethod(oc, methodName, pcs);
            Object o1 = oc.newInstance();  // add by chf
            returnValue = m.invoke(o1, parameters);
            if (key != null && returnValue != null) {
                addCacheData(key, returnValue, false);
            }
            if (returnValue == null) {
                returnValue = m.getReturnType();
            }
        } catch(Exception e) {
            log.error("调用方法失败,methodName=" + methodName);
            if (key != null) {
                removeCacheData(key);
                log.error("更新缓存失败,缓存key=" + key);
            }
            e.printStackTrace();
        }
        return returnValue;
    }

    /**
     * 找不到完全匹配的方法时,对参数进行向父类匹配
     * 因为方法aa(java.util.List) 与 aa(java.util.ArrayList)不能自动匹配到
     *
     * @param oc
     * @param methodName
     * @param pcs
     * @return
     * @throws NoSuchMethodException
     * @throws NoSuchMethodException
     */
    private Method matchMethod(Class oc, String methodName, Class[] pcs
            ) throws NoSuchMethodException, SecurityException {
        try {
            Method method = oc.getDeclaredMethod(methodName, pcs);
            return method;
        } catch (NoSuchMethodException e) {
            Method[] ms = oc.getDeclaredMethods();
            aa:for (int i = 0; i < ms.length; i++) {
                if (ms[i].getName().equals(methodName)) {
                    Class[] pts = ms[i].getParameterTypes();
                    if (pts.length == pcs.length) {
                        for (int j = 0; j < pts.length; j++) {
                            if (!pts[j].isAssignableFrom(pcs[j])) {
                                break aa;
                            }
                        }
                        return ms[i];
                    }
                }
            }
            throw new NoSuchMethodException();
        }
    }

    /**
     * 新启线程后台调用给定方法更新缓存中数据据
     * @param o
     * @param methodName
     * @param parameters
     * @param key
     */
    private void daemonInvoke(Object o, String methodName,Object[] parameters, String key) {
        if (!threadKeys.contains(key)) {
            InvokeThread t = new InvokeThread(o, methodName, parameters, key);
            t.start();
        }
    }

 

   

 /**
     * 些类存放方法的主调对像,名称及参数数组
     * @author hiya
     *
     */
    public class MethodInfo {
        private Object o;
        private String methodName;
        private Object[] parameters;
        public MethodInfo(Object o, String methodName,Object[] parameters) {
            this.o = o;
            this.methodName = methodName;
            this.parameters = parameters;
        }
        public String getMethodName() {
            return methodName;
        }
        public void setMethodName(String methodName) {
            this.methodName = methodName;
        }
        public Object getO() {
            return o;
        }
        public void setO(Object o) {
            this.o = o;
        }
        public Object[] getParameters() {
            return parameters;
        }
        public void setParameters(Object[] parameters) {
            this.parameters = parameters;
        }

        public String toString() {
            StringBuffer str = new StringBuffer(methodName);
            if (parameters != null) {
                str.append("(");
                for (int i = 0; i < parameters.length; i++) {
                    if (parameters[i] instanceof Object[]) {
                        str.append(Arrays.toString((Object[])parameters[i])).append(",");
                    } else {
                        str.append(parameters[i]).append(",");
                    }
                }
                str.append(")");
            }
            return str.toString();
        }
    }

 

    /**
     * 线程调用方法
     * @author hiya
     *
     */
    private class InvokeThread extends Thread {
        private Object o;
        private String methodName;
        private Object[] parameters;
        private String key;
        public InvokeThread(Object o, String methodName,Object[] parameters, String key) {
            this.o = o;
            this.methodName = methodName;
            this.parameters = parameters;
            this.key = key;
        }

        public void run() {
            threadKeys.add(key);
            invoke(o, methodName, parameters, key);
            threadKeys.remove(key);
        }
    }

    /**
     * 移除缓存中的数据
     * @param key
     */
    public void removeCacheData(String key) {
        cacheMap.remove(key);
    }

    /**
     * 移除所有缓存中的数据
     *
     */
    public void removeAllCacheData() {
        cacheMap.clear();
    }

    public String toString() {
        StringBuffer sb = new StringBuffer("************************ ");
        sb.append("正在更新的缓存数据: ");
        for (int i = 0; i < threadKeys.size(); i++) {
            sb.append(threadKeys.get(i)).append(" ");
        }
        sb.append("当前缓存大小:").append(cacheMap.size()).append(" ");
        sb.append("************************");
        return sb.toString();
    }

}

 

 实际使用:

package atest.hiyaCache;

//import javax.swing.text.Document;

import atest.hiyaCache.CacheOperation.MethodInfo;

public class CacheExec {

    /**
     * @param args
     */
    public static void main(String[] args) {//        String province = request.getParameter("province");
//        String city= request.getParameter("city");
//        String county= request.getParameter("county");
//        Document doc = XMLBuilder.buildLatelyKeyword(kwm.latelyKeyword(province, city, county));
//        out.write(doc);

        String province = "Jiangsu ";
        String city= "Nanjing ";
        String county= "Jiangning";
        CacheOperation co = CacheOperation.getInstance();
//        MethodInfo mi = co.new MethodInfo(kwm, "latelyKeyword", new Object[]{province, city, county});
//        Document doc = (Document )co.getCacheData(XMLBuilder.class,"buildLatelyKeyword",new Object[]{mi}, 120000, 0);
//        out.write(doc);
       while (true){
            // [email protected] 
            MethodInfo mi = co.new MethodInfo(AddrDetail.class, "latelyKeyword", new Object[]{province, city, county});
            // 120000 毫秒(2分钟)更新缓存
            String aa = (String)co.getCacheData(AddrDetail.class,"buildCache",new Object[]{mi}, 120000, 5);
            System.out.println("CacheExec:main=" + aa);
        }    }

}

 


当缓存次数超过5时,重新执行。看一下执行结果。

AddrDetail.latelyKeyword=Jiangsu Nanjing Jiangning
AddrDetail.buildCache=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
AddrDetail.latelyKeyword=Jiangsu Nanjing Jiangning
AddrDetail.buildCache=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
CacheExec:main=Jiangsu Nanjing Jiangning
AddrDetail.latelyKeyword=Jiangsu Nanjing Jiangning
AddrDetail.buildCache=Jiangsu Nanjing Jiangning

   大数据量的缓存肯定还是要保存到文件,这个时候光用JDK来实现就比较复杂了,这里当然要用到第三方开源框架来实现了,常用的有Oscache,Ehcache,Jcache,Jbosscache等等很多,推荐还是Ehcache与Oscache,hibernate采用了Ehcache做为其缓存机制默认实现,Oscache没深入研究,但是Ibatis推荐的缓存为这个。


   为什么要缓存,无非就是节省访问时间已经大并发量带来的访问上资源的消耗,这个资源有软资源和硬资源,做java的几乎每天都在有意识或者无意识的跟缓存打交道,最常见的缓存,Connection Pool,还有IOC其实也算一个。
       

  缓存用的好能提高性能,用的不好反而会急剧的降低产品的性能,就拿hibernate来说,hibernate性能肯定不如jdbc,但是缓存用的好的话增删改查的性能相差无几,hibernate缓存最核心的部分个人觉得在于对象的有效性,缓存的命中率越高意味着性能越高,命中率跟缓存对象的有效性息息相关,如何保证对象有效这个很难,也很有搞头,如果缓存中对象有效性很差,其性能甚至会低于不用缓存,因为缓存本身就会耗性能跟资源,缓存的对象很多都很快失效了无疑得不偿失,还有缓存的深度也有讲究,这个深度是指从页面到数据库,显然是页面缓存的性能最好,因为调用页面缓存消耗的资源最少,当然现实中是不可能有太多页面缓存的。



 

郑重声明:本站内容如果来自互联网及其他传播媒体,其版权均属原媒体及文章作者所有。转载目的在于传递更多信息及用于网络分享,并不代表本站赞同其观点和对其真实性负责,也不构成任何其他建议。