You need to present common data to many clients, and you want to cache this common data in a data structure of limited size, evicting the least recently used entry when the maximum size limit has been reached.

Use LRUMap—a fixed size map that uses the least recently used algorithm to evict entries once the maximum size has been reached. The least recently used algorithm evicts the element that has not been used for the longest amount of time. The following example demonstrates the use of an LRUMap to store stock quotes:

import java.util.Map;
import org.apache.commons.collections.map.LRUMap;

cache = new LRUMap( 5 );
       
// Populate the cache with 5 stock prices
cache.put( "MSFT", new Float( 0.03 ) );
cache.put( "TSC", new Float( 0.001 ) );
cache.put( "LU", new Float( 23.30 ) );
cache.put( "CSCO", new Float( 242.20 ) );
cache.put( "P", new Float( 10.23 ) );
        
// Now use some of the entries in the cache
Float cscoPrice  = (Float) cache.get( "CSCO" );
Float msPrice = (Float) cache.get( "MSFT" );
Float tscPrice = (Float) cache.get( "TSC" );
Float luPrice = (Float) cache.get( "LU" );
Float pPrice = (Float) cache.get( "P" );
Float msPrice2 = (Float) cache.get( "MSFT" );
       
// Add another price to the Map, this should kick out the LRU item.
               cache.put( "AA", new Float( 203.20 ) );
                    
               // CSCO was the first price requested, it is therefore the
               // least recently used.
               if( !cache.containsKey("CSCO") ) {
                   System.out.println( "As expected CSCO was discarded" );
               }

Since the LRUMap created in this example can only hold five elements, the “CSCO” quote is evicted when “AA” is added to the map because it is the “least recently used.”

An LRUMap is appropriate when a system presents common data to a number of clients. For example, if a web site displays breaking news stories, objects representing these stories can be stored in a size-limited cache with an LRU eviction policy, such as LRUMap. Demand for objects that represent breaking news stories can be very time sensitive; many customers read the same story at the same time. Your application server has a size-limited LRUMap that holds 100 news stories, and, when a new story is added, the LRUMap evicts the stalest story in the Map.

The LRUMap is an ideal solution for a cache, but, if your application needs to make heavy use of caching, take a look at the following open source cache implementations—OSCache or JCS. For more information about OSCache, see the OpenSymphony OSCache project page (http://www.opensymphony.com/oscache/). The Java Caching System (JCS) was an offshoot of the Jakarta Turbine project; for more information about JCS, see the JCS project page (http://jakarta.apache.org/turbine/jcs/).