Barrier synchronization seemingly can be improved by applying tree-based dissemination structure[1].
[1]. Practical Barrier Synchronisation. ftp://ftp.comlab.ox.ac.uk/pub/Documents/techpapers/Jonathan.Hill/HillSkill_barrier.ps.Z -----Original message----- From:Edward J. Yoon <[email protected]> To:[email protected] Date:Mon, 4 Jul 2011 09:13:18 +0900 Subject:Our TODO things Hi all, Here's our TODO things: 1. Improve lock and barrier synchronization 2. Support configurable task slot 3. Add Pregel-like graph computing framework based on BSP engine 4. Support checkpoint based on HDFS 5. Support Fault Tolerant based on checkpoint 6. Support Input/OutputFormatter like MapReduce 7. Support compatible job with MapReduce2 8. Support send and receive large messages or many messages 9. Support message compression 10. Advanced job scheduler * Performance and Fault Tolerant issues requires a lot of experiences and tests. -- Best Regards, Edward J. Yoon @eddieyoon -- ChiaHung Lin Department of Information Management National University of Kaohsiung Taiwan
