java - 线程安全映射 : improve performance

Question

public class EntityUtils
{
    private static final Map<String, Map<String, String>> searchMap = new HashMap<>();

    private static Map<String, String> getSearchablePathMap(String key)
    {
        synchronized(searchMap)
        {
            Map<String, String> pathMap = searchMap.get(key);
            if(pathMap != null) return pathMap;

            pathMap = new HashMap<>();
            pathMap.put(..., ...);
            ...
            // heavy map population operations
            ...

            pathMap = Collections.unmodifiableMap(pathMap);

            searchMap.put(key, pathMap);
        }

        return pathMap;
    }
}

map 条目永远不会被删除。

由于人口众多，无法使用 ConcurrentHashMap.putIfAbsent()。

当 key 已经存在时，您能否建议性能改进以避免同步？

Answer 1

无法避免 get 上的某种级别的同步，但可以允许并发读取。考虑使用 ReadWriteLock保护对您的搜索 map 的访问。这仅适用于获取远多于放置的情况。

public class EntityUtils
{
    private static final ReadWriteLock RW_LOCK = new ReentrantReadWriteLock();
    private static final Map<String, Map<String, String>> searchMap = new HashMap<>();

    private static Map<String, String> getSearchablePathMap(String key)
    {
        RW_LOCK.readLock().lock();
        try
        {
            Map<String, String> pathMap = searchMap.get(key);
            if(pathMap != null) return pathMap;
        }
        finally
        {
            RW_LOCK.readLock().unlock();
        }

        RW_LOCK.writeLock().lock();
        try
        {
            //first check to see if a previous holder of write lock built map for us
            Map<String, String> pathMap = searchMap.get(key);
            if(pathMap != null) return pathMap;

            pathMap = new HashMap<>();
            pathMap.put(..., ...);
            ...
            // heavy map population operations
            ...

            pathMap = Collections.unmodifiableMap(pathMap);

            searchMap.put(key, pathMap);
        }
        finally
        {
            RW_LOCK.writeLock().unlock();
        }

        return pathMap;
    }
}