Question

I need to get one data from a collection (get() calls in the range of 100K for a single file processing).

public class DemoCollection {

    private Map<GroupCriteria, GroupData> collectionHolder = new ConcurrentHashMap<GroupCriteria, GroupData>();

    /**
     * 
     * @param groupCriteria
     *            GroupCriteria
     * @return GroupData
     */
    public GroupData getGroupForGroupingCriteriaOne(GroupCriteria groupCriteria) {
        GroupData groupData = null;
        if (collectionHolder.containsKey(groupCriteria)) {
            groupData = collectionHolder.get(groupCriteria);
        } else {
            // Get from database
        }
        return groupData;
    }

    /**
     * 
     * @param groupCriteria
     *            GroupCriteria
     * @return GroupData
     */
    public GroupData getGroupForGroupingCriteriaTwo(GroupCriteria groupCriteria) {
        GroupData groupData = null;
        if ((groupData = collectionHolder.get(groupCriteria)) == null) {
            // GEt from database
        }
        return groupData;
    }
}

Which is the best practice in this regard ? The approach one (getGroupForGroupingCriteriaOne) , two (getGroupForGroupingCriteriaTwo) or neither ?

Usually I ignore these premature optimization things, but since the get() calls are too huge I’m little bothered.

Could you please advice ?

Answer 1

Consider using Guava’s MapMaker:

private ConcurrentMap<GroupCriteria, GroupData> collectionHolder = new MapMaker()
        .makeComputingMap(
           new Function<GroupCriteria, GroupData>() {
              @Override
              public GroupData apply(GroupCriteria key) {
                 //get from database and return
              }
        });

This ConcurrentMap will handle all concurrent requests for you. See the MapMaker documentation for a list of all configurable features of the produced map.