`
lspgdut
  • 浏览: 199007 次
  • 性别: Icon_minigender_1
  • 来自: 广州
文章分类
社区版块
存档分类
最新评论

cache

阅读更多
/**
* $RCSfile: Cache.java,v $
* $Revision: 1.5 $
* $Date: 2002/05/10 21:53:20 $
*
* Copyright (C) 1999-2002 CoolServlets, Inc. All rights reserved.
*
* This software is the proprietary information of CoolServlets, Inc.
* Use is subject to license terms.
*/

package com.jivesoftware.util;

/**
*  cache的一般目的.它可以通过快速访问内存中标有唯一标记的对象集合!
* 所有的键和值添加到cache中必须实现 Serializable这个接口.
*  值可以实现Cacheable这个接口, 在cache里边可以更快的决定对象的大小.
* These restrictions allow a cache to never grow larger than a specified number
* of bytes and to optionally be distributed over a cluster of servers.<p>
*
* If the cache does grow too large, objects will be removed such that those
* that are accessed least frequently are removed first. Because expiration
* happens automatically, the cache makes <b>no</b> gaurantee as to how long
* an object will remain in cache after it is put in.<p>
*
* Optionally, a maximum lifetime for all objects can be specified. In that
* case, objects will be deleted from cache after that amount of time, even
* if they are frequently accessed. This feature is useful if objects put in
* cache represent data that should be periodically refreshed; for example,
* information from a database.<p>
*
* All cache operations are thread safe.<p>
*
* @see Cacheable
* @author Matt Tucker
*/
public interface Cache extends java.util.Map {

    /**
     * 返回这个cache的名字
     *
     * @return the name of the cache.
     */
    String getName();

    /**
     * 返回cache的最大使用字节数.如果这个cache值大于最大值,我们就会把使用频率最少的记录删除
     *
     * @return the maximum size of the cache in bytes.
     */
    int getMaxCacheSize();

    /**
     * 设置cache的最大值字节. If the cache grows
     * larger than the max size, the least frequently used items will be removed.
     *
     * @param maxSize the maximum size of the cache in bytes.
     */
    void setMaxCacheSize(int maxSize);

    /**
     * Returns the maximum number of milleseconds that any object can live
     * in cache. Once the specified number of milleseconds passes, the object
     * will be automatically expried from cache. If the max lifetime is set
     * to -1, then objects never expire.
     *
     * @return the maximum number of milleseconds before objects are expired.
     */
    long getMaxLifetime();

    /**
     * Sets the maximum number of milleseconds that any object can live
     * in cache. Once the specified number of milleseconds passes, the object
     * will be automatically expried from cache. If the max lifetime is set
     * to -1, then objects never expire.
     *
     * @param maxLifetime the maximum number of milleseconds before objects are expired.
     */
    void setMaxLifetime(long maxLifetime);

    /**
     * Returns the size of the cache contents in bytes. This value is only a
     * rough approximation, so cache users should expect that actual VM
     * memory used by the cache could be significantly higher than the value
     * reported by this method.
     *
     * @return the size of the cache contents in bytes.
     */
    int getCacheSize();

    /**
     * Returns the number of cache hits. A cache hit occurs every
     * time the get method is called and the cache contains the requested
     * object.<p>
     *
     * Keeping track of cache hits and misses lets one measure how efficient
     * the cache is; the higher the percentage of hits, the more efficient.
     *
     * @return the number of cache hits.
     */
    long getCacheHits();

    /**
     * Returns the number of cache misses. A cache miss occurs every
     * time the get method is called and the cache does not contain the
     * requested object.<p>
     *
     * Keeping track of cache hits and misses lets one measure how efficient
     * the cache is; the higher the percentage of hits, the more efficient.
     *
     * @return the number of cache hits.
     */
    long getCacheMisses();
}














**
* $RCSfile: DefaultCache.java,v $
* $Revision: 1.9 $
* $Date: 2002/07/15 13:19:01 $
*
* Copyright (C) 1999-2001 CoolServlets, Inc. All rights reserved.
*
* This software is the proprietary information of CoolServlets, Inc.
* Use is subject to license terms.
*/
package com.jivesoftware.util;

import java.util.*;
import java.io.*;

/**
* Default, non-distributed implementation of the Cache interface.
* The algorithm for cache is as follows: a HashMap is maintained for fast
* object lookup. Two linked lists are maintained: one keeps objects in the
* order they are accessed from cache, the other keeps objects in the order
* they were originally added to cache. When objects are added to cache, they
* are first wrapped by a CacheObject which maintains the following pieces
* of information:<ul>
*    <li> The size of the object (in bytes).
*    <li> A pointer to the node in the linked list that maintains accessed
*         order for the object. Keeping a reference to the node lets us avoid
*         linear scans of the linked list.
*    <li> A pointer to the node in the linked list that maintains the age
*         of the object in cache. Keeping a reference to the node lets us avoid
*         linear scans of the linked list.</ul>
*
* To get an object from cache, a hash lookup is performed to get a reference
* to the CacheObject that wraps the real object we are looking for.
* The object is subsequently moved to the front of the accessed linked list
* and any necessary cache cleanups are performed. Cache deletion and expiration
* is performed as needed.
*
* @author Matt Tucker
*/
public class DefaultCache implements Cache {

    /**
     * The map the keys and values are stored in.
     */
    protected Map map;

    /**
     * Linked list to maintain order that cache objects are accessed
     * in, most used to least used.
     */
    protected LinkedList lastAccessedList;

    /**
     * Linked list to maintain time that cache objects were initially added
     * to the cache, most recently added to oldest added.
     */
    protected LinkedList ageList;

   /**
    * Maximum size in bytes that the cache can grow to.
    */
    private int maxCacheSize;

    /**
     * Maintains the current size of the cache in bytes.
     */
    private int cacheSize = 0;

    /**
     * Maximum length of time objects can exist in cache before expiring.
     */
    protected long maxLifetime;

    /**
     * Maintain the number of cache hits and misses. A cache hit occurs every
     * time the get method is called and the cache contains the requested
     * object. A cache miss represents the opposite occurence.<p>
     *
     * Keeping track of cache hits and misses lets one measure how efficient
     * the cache is; the higher the percentage of hits, the more efficient.
     */
    protected long cacheHits, cacheMisses = 0L;

    /**
     * The name of the cache.
     */
    private String name;

    /**
     * Create a new cache and specify the maximum size of for the cache in
     * bytes, and the maximum lifetime of objects.
     *
     * @param name a name for the cache.
     * @param maxSize the maximum size of the cache in bytes.
     * @param maxLifetime the maximum amount of time objects can exist in
     *    cache before being deleted. -1 means objects never expire.
     */
    protected DefaultCache(String name, int maxSize, long maxLifetime) {
        this.name = name;
        this.maxCacheSize = maxSize;
        this.maxLifetime = maxLifetime;

        // Our primary data structure is a hash map. The default capacity of 11
        // is too small in almost all cases, so we set it bigger.
        map = new HashMap(103);

        lastAccessedList = new LinkedList();
        ageList = new LinkedList();
    }

    public synchronized Object put(Object key, Object value) {
        // Delete an old entry if it exists.
        remove(key);

        int objectSize = calculateSize(value);

        // If the object is bigger than the entire cache, simply don't add it.
        if (objectSize > maxCacheSize * .90) {
            System.err.println("Cache: " + name + " -- object with key " + key +
                    " is too large to fit in cache. Size is " + objectSize);
            return value;
        }
        cacheSize += objectSize;
        CacheObject cacheObject = new CacheObject(value, objectSize);
        map.put(key, cacheObject);
        // Make an entry into the cache order list.
        LinkedListNode lastAccessedNode = lastAccessedList.addFirst(key);
        // Store the cache order list entry so that we can get back to it
        // during later lookups.
        cacheObject.lastAccessedListNode = lastAccessedNode;
        // Add the object to the age list
        LinkedListNode ageNode = ageList.addFirst(key);
        // We make an explicit call to currentTimeMillis() so that total accuracy
        // of lifetime calculations is better than one second.
        ageNode.timestamp = System.currentTimeMillis();
        cacheObject.ageListNode = ageNode;

        // If cache is too full, remove least used cache entries until it is
        // not too full.
        cullCache();

        return value;
    }

    public synchronized Object get(Object key) {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        CacheObject cacheObject = (CacheObject)map.get(key);
        if (cacheObject == null) {
            // The object didn't exist in cache, so increment cache misses.
            cacheMisses++;
            return null;
        }

        // The object exists in cache, so increment cache hits. Also, increment
        // the object's read count.
        cacheHits++;
        cacheObject.readCount++;

        // Remove the object from it's current place in the cache order list,
        // and re-insert it at the front of the list.
        cacheObject.lastAccessedListNode.remove();
        lastAccessedList.addFirst(cacheObject.lastAccessedListNode);

        return cacheObject.object;
    }

    public synchronized Object remove(Object key) {
        CacheObject cacheObject = (CacheObject)map.get(key);
        // If the object is not in cache, stop trying to remove it.
        if (cacheObject == null) {
            return null;
        }
        // remove from the hash map
        map.remove(key);
        // remove from the cache order list
        cacheObject.lastAccessedListNode.remove();
        cacheObject.ageListNode.remove();
        // remove references to linked list nodes
        cacheObject.ageListNode = null;
        cacheObject.lastAccessedListNode = null;
        // removed the object, so subtract its size from the total.
        cacheSize -= cacheObject.size;
        return cacheObject.object;
    }

    public synchronized void clear() {
        Object [] keys = map.keySet().toArray();
        for (int i=0; i<keys.length; i++) {
            remove(keys[i]);
        }

        // Now, reset all containers.
        map.clear();
        lastAccessedList.clear();
        lastAccessedList = new LinkedList();
        ageList.clear();
        ageList = new LinkedList();

        cacheSize = 0;
        cacheHits = 0;
        cacheMisses = 0;
    }

    public int size() {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        return map.size();
    }

    public boolean isEmpty() {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        return map.isEmpty();
    }

    public Collection values() {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        Object [] cacheObjects = map.values().toArray();
        Object [] values = new Object[cacheObjects.length];
        for (int i=0; i<cacheObjects.length; i++) {
            values[i] = ((CacheObject)cacheObjects[i]).object;
        }
        return Collections.unmodifiableList(Arrays.asList(values));
    }

    public boolean containsKey(Object key) {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        return map.containsKey(key);
    }

    public void putAll(Map map) {
        for (Iterator i=map.keySet().iterator(); i.hasNext(); ) {
            Object key = i.next();
            Object value = map.get(key);
            put(key, value);
        }
    }

    public boolean containsValue(Object value) {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        int objectSize = calculateSize(value);
        CacheObject cacheObject = new CacheObject(value, objectSize);
        return map.containsValue(cacheObject);
    }

    public Set entrySet() {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        return Collections.unmodifiableSet(map.entrySet());
    }

    public String getName() {
        return name;
    }

    public Set keySet() {
        // First, clear all entries that have been in cache longer than the
        // maximum defined age.
        deleteExpiredEntries();

        return Collections.unmodifiableSet(map.keySet());
    }

    public long getCacheHits() {
        return cacheHits;
    }

    public long getCacheMisses() {
        return cacheMisses;
    }

    public int getCacheSize() {
        return cacheSize;
    }

    public int getMaxCacheSize() {
        return maxCacheSize;
    }

    public void setMaxCacheSize(int maxCacheSize) {
        this.maxCacheSize = maxCacheSize;
        // It's possible that the new max size is smaller than our current cache
        // size. If so, we need to delete infrequently used items.
        cullCache();
    }

    public long getMaxLifetime() {
        return maxLifetime;
    }

    public void setMaxLifetime(long maxLifetime) {
        this.maxLifetime = maxLifetime;
    }

    /**
     * Returns the size of an object in bytes. Determining size by serialization
     * is only used as a last resort.
     *
     * @return the size of an object in bytes.
     */
    private int calculateSize(Object object) {
        // If the object is Cacheable, ask it its size.
        if (object instanceof Cacheable) {
            return ((Cacheable)object).getCachedSize();
        }
        // Coherence puts DataInputStream objects in cache.
        else if (object instanceof java.io.DataInputStream) {
            int size = 1;
            try {
                size = ((DataInputStream)object).available();
            }
            catch (IOException ioe) { }
            return size;
        }
        // Check for other common types of objects put into cache.
        else if (object instanceof Long) {
            return CacheSizes.sizeOfLong();
        }
        else if (object instanceof Integer) {
            return CacheSizes.sizeOfObject() + CacheSizes.sizeOfInt();
        }
        else if (object instanceof Boolean) {
            return CacheSizes.sizeOfObject() + CacheSizes.sizeOfBoolean();
        }
        else if (object instanceof long []) {
            long [] array = (long [])object;
            return CacheSizes.sizeOfObject() + array.length * CacheSizes.sizeOfLong();
        }
        // Default behavior -- serialize the object to determine its size.
        else {
            int size = 1;
            try {
                // Default to serializing the object out to determine size.
                NullOutputStream out = new NullOutputStream();
                ObjectOutputStream outObj = new ObjectOutputStream(out);
                outObj.writeObject(object);
                size = out.size();
            }
            catch (IOException ioe) {
                ioe.printStackTrace();
            }
            return size;
        }
    }

    /**
     * Clears all entries out of cache where the entries are older than the
     * maximum defined age.
     */
    protected synchronized void deleteExpiredEntries() {
        // Check if expiration is turned on.
        if (maxLifetime <= 0) {
            return;
        }

        // Remove all old entries. To do this, we remove objects from the end
        // of the linked list until they are no longer too old. We get to avoid
        // any hash lookups or looking at any more objects than is strictly
        // neccessary.
        LinkedListNode node = ageList.getLast();
        // If there are no entries in the age list, return.
        if (node == null) {
            return;
        }

        // Determine the expireTime, which is the moment in time that elements
        // should expire from cache. Then, we can do an easy to check to see
        // if the expire time is greater than the expire time.
        long expireTime = CacheFactory.currentTime - maxLifetime;

        while(expireTime > node.timestamp) {
            // Remove the object
            remove(node.object);

            // Get the next node.
            node = ageList.getLast();
            // If there are no more entries in the age list, return.
            if (node == null) {
                return;
            }
        }
    }

    /**
     * Removes objects from cache if the cache is too full. "Too full" is
     * defined as within 3% of the maximum cache size. Whenever the cache is
     * is too big, the least frequently used elements are deleted until the
     * cache is at least 10% empty.
     */
    protected final void cullCache() {
        // See if the cache size is within 3% of being too big. If so, clean out
        // cache until it's 10% free.
        if (cacheSize >= maxCacheSize * .97) {
            // First, delete any old entries to see how much memory that frees.
            deleteExpiredEntries();
            int desiredSize = (int)(maxCacheSize * .90);
            while (cacheSize > desiredSize) {
                // Get the key and invoke the remove method on it.
                remove(lastAccessedList.getLast().object);
            }
        }
    }

    /**
     * An extension of OutputStream that does nothing but calculate the number
     * of bytes written through it.
     */
    private static class NullOutputStream extends OutputStream {

        int size = 0;

        public void write(int b) throws IOException  {
            size++;
        }

        public void write(byte[] b) throws IOException {
            size += b.length;
        }

        public void write(byte[] b, int off, int len) {
            size += len;
        }

        /**
         * Returns the number of bytes written out through the stream.
         *
         * @return the number of bytes written to the stream.
         */
        public int size() {
            return size;
        }
    }
}

分享到:
评论

相关推荐

    cache性能分析实验

    ### Cache性能分析实验知识点 #### 实验背景与目标 本实验旨在通过使用SimpleScalar模拟器对Cache性能进行深入分析,以此来加深对Cache基础知识、结构及其工作原理的理解。此外,还将探讨并量化Cache的主要参数...

    springboot整合jetcache完整代码

    JetCache是阿里巴巴开源的一款高性能、轻量级的分布式缓存框架,适用于微服务架构中的缓存场景。本文将详细介绍如何在SpringBoot项目中整合JetCache,并探讨其本地缓存和Redis缓存的使用,以及如何实现多缓存的并行...

    Oracle Buffer和Cache的区别

    Oracle数据库中的Buffer Cache和一般的Cache概念虽然相似,但它们在具体应用中有着不同的侧重点。首先,我们需要理解Buffer Cache的基本概念。在Oracle数据库系统中,Buffer Cache是内存结构的一部分,它存储了最近...

    java连接cache数据库说明,数据库驱动,cache可视化工具

    Java连接Cache数据库主要涉及到的是如何使用Java编程语言与Intersystems Cache数据库进行交互。Intersystems Cache是一款高性能、面向对象的数据库系统,广泛应用于医疗、金融等领域的复杂数据管理。在Java环境中,...

    PrimoCache重置工具

    《PrimoCache重置工具详解及应用》 PrimoCache是一款高效、实用的硬盘缓存软件,它通过在系统内存中创建虚拟缓存,显著提升硬盘读写性能,从而优化系统运行速度。然而,如同大多数试用软件一样,PrimoCache在一定...

    cache性能分析实验报告.docx

    【Cache性能分析实验报告】 本实验旨在深入理解Cache的基本概念、结构和工作原理,通过实际操作分析Cache的容量、相联度和块大小对性能的影响,以及不同替换算法的效果。实验采用Vmware虚拟机上的Redhat 9.0 Linux...

    高速缓存(Cache)的Verilog代码

    该工程包含数据缓存D_Cache和指令缓存I_Cache的Verilog代码和仿真文件,Cache的详细技术参数包含在.v文件的注释中。 直接相连16KB D_Cache Cache写策略: 写回法+写分配 (二路)组相连16KB I_Cache Cache替换策略: ...

    TIC6678多核编程Cache总结.pdf

    TI-6678 DSP多核编程中的Cache技术是一个复杂而重要的议题,尤其在多核处理环境下。Cache是计算机架构中用于减少处理器访问内存所需平均时间的一个高速数据存储区域。它作为处理器和主存储器之间的临时缓存,能够极...

    分块矩阵优化cache

    ### 分块矩阵优化Cache:深度解析与应用策略 #### 核心知识点概览: 1. **Cache基础原理**:理解高速缓存(Cache)在现代计算机系统中的关键作用及其内部结构,包括标记存储器和数据存储器的功能。 2. **Cache失效...

    Cache的工作原理

    ### Cache的工作原理详解 #### 一、引言 在现代计算机体系结构中,为了提高处理器访问数据的速度,引入了多种缓存技术。其中,Cache作为连接CPU与主存的重要环节,其工作原理对于理解计算机系统性能至关重要。本文...

    计算机体系结构cache实验报告

    在本“计算机体系结构cache实验报告”中,我们主要探讨了Cache存储过程的模拟和性能分析,重点关注了不同因素如关联方式、Cache容量、关联度和块大小对Cache性能的影响。实验采用控制变量法,通过操作系统试验中的...

    PrimoCache v3.09.zip

    《PrimoCache v3.09:硬盘缓存利器的深度解析》 PrimoCache,一个在IT领域中被广泛使用的高效硬盘缓存工具,其最新版本v3.09,为用户带来了更为优化的存储性能提升。这款软件的核心功能在于通过在内存中创建临时...

    cache-api-1.1.1-API文档-中文版.zip

    赠送jar包:cache-api-1.1.1.jar; 赠送原API文档:cache-api-1.1.1-javadoc.jar; 赠送源代码:cache-api-1.1.1-sources.jar; 赠送Maven依赖信息文件:cache-api-1.1.1.pom; 包含翻译后的API文档:cache-api-...

    cache_axi4.rar

    在计算机系统中,Cache是一种高速数据存储部件,用于暂时存储CPU频繁访问的内存数据,以减少主内存与CPU之间的数据传输延迟,提高系统的整体性能。本文将深入探讨基于AXI4(Advanced eXtensible Interface)总线协议...

    logisim及全相联cache设计.rar

    全相联Cache( Fully-Associative Cache)是Cache组织方式的一种,与直接映射Cache和组相联Cache不同,它的每一个块都可以映射到Cache的任何一个位置上,这提供了更大的灵活性,但也带来了更高的复杂性。 全相联...

    jdbc连接cache的demo及jar包,自己备份.rar

    标题"jdbc连接cache的demo及jar包,自己备份.rar"指出这是一个关于使用JDBC(Java Database Connectivity)连接Cache数据库的示例项目,其中包含了必要的jar包,并且用户已经将其作为个人备份保存。这里的“Cache”...

    Guava-Cache本地缓存案例代码

    Guava Cache是Google Guava库中的一个强大特性,它提供了高效的本地缓存解决方案,用于存储经常访问的数据,以减少对远程服务或计算的调用,从而提高应用性能。本案例代码将详细介绍Guava Cache的使用,包括缓存的...

    nginx_cache_purge.zip

    《Nginx Cache Purge:高效管理Web缓存的利器》 在当今互联网环境中,Web服务器的性能优化至关重要,而缓存技术则是其中的关键一环。Nginx,以其高性能、稳定性以及模块化的特性,成为了许多网站首选的反向代理和...

    Linux驱动中的DMA和Cache一致性问题

    然而,DMA和Cache之间存在一致性问题,特别是在某些嵌入式平台上,DMA操作可能会绕过Cache,导致数据不一致,这就需要通过一系列策略来确保Cache一致性。 在DMA机制中,有两类主要的Cache一致性问题:流式DMA...

    实验3 直接相联Cache设计1

    计算机组成原理实验指导书中的“实验3 直接相联Cache设计”着重讲解了Cache的基础知识,特别是直接相联Cache的结构、设计方法以及其实现。以下是对实验内容的详细阐述: 1. **直接相联Cache的基本结构**: - **...

Global site tag (gtag.js) - Google Analytics