how to search and return the first non null result in a distributed cache

[matlach]

Hello everyone,

I was wondering if there was any efficient way to search and return the first non null result in a distributed cache. For now I have used the Map-Reduce framework to perform a basic search but I do not think this is quite optimal (well it is not !).

Here is my example :

    public class Session {
        private String userId;
    }

    public static class GetSessionByUserIdMapper implements Mapper<String, Session, String, Session>{
        private static final long serialVersionUID = 1L;

        private String userId;
        
        public GetSessionByUserIdMapper(String userId){
            this.userId = userId;
        }
        
        @Override
        public void map(String key, Session value, Collector<String, Session> collector) {
            if (value.getUserId().equals(userId)){
                collector.emit(value.getUserId(), value);
                // at that moment we would be done
                // is it possible to stop the map reduce operation and return this value ?
            }
        }
    }
    
    public static class GetSessionByUserIdReducer implements Reducer<String, Session>{
        private static final long serialVersionUID = 1L;

        @Override
        public Session reduce(String reducedKey, Iterator<Session> iter) {
            if (! iter.hasNext()){
                return null;
            }
            return iter.next();
        }
    }

    public static class GetSessionByUserIdCollator implements Collator<String, Session, Session>{
        @Override
        public Session collate(Map<String, Session> reducedResults)
        {
            Iterator<Session> iterator = reducedResults.values().iterator();
            if (iterator.hasNext()){
                return iterator.next();
            }
            return null;
        }
    }

That is a lot of operations to perform this quite simple task.

I'm aware of the existence of the Infinispan-HibernateSearch-Lucene bridge and I know it would technically work, though I would really like to far from that approach.

Is there any other way to perform this task ?

Big thanks,