It's that time again, thinking about the next release of ZK. From what I'm hearing people seem interested in improving scalability, manageability, and continue to improve client support.

In particular I believe we should look at the following going forward. Of course ZooKeeper is open to submissions in that aren't on this list. If you have any suggestions please feel free to enter a JIRA, submit a patch, or comment on this thread.
http://wiki.apache.org/hadoop/ZooKeeper/HowToContribute

3.3 (~4 months out)
  Observer support - ZOOKEEPER-368

  Allow dynamic changes to ensemble w/o requiring restart - ZOOKEEPER-107

  Move from direct NIO to framework (grizzly?)
    benefit of built-in ssl support for connections

  optimize session tracking
    no expiration of session if client has no ephemerals
    huge scalability win if we can do this (for a very common use case)

  hide connection loss from client - ZOOKEEPER-22

  optimize server
     profiling for hotspots
replace String use with binary buffer - elim conversion overhead (server only)

  expand system tests
    ensure b/w compat between current and previous version
move some unit tests into system test framework (esp large quorum tests)
    integrate performance/benchmark tests
  cleanup the c binding
    logging and general code cleanliness
  review jmx object naming scheme
     error handling, general cruft
  move examples out of docs into contrib/examples (like hadoop core)
    add more examples
  build changes
    ivy? push jars to maven repo?
    split jars? common, server, client, test

  horizontally partitioned zk
  support multiple client protocols/marshalling
    (hadoop rpc?)
    AVRO is in the process of it's first release
  quota support for limiting (rather than just reporting)

4.0 "API/data breakage release" (q1 2010)
  API changes
move all "counter" fields from int -> long (version numbers for example) requires migration of snapshots (ie persistence/marshalling will change)
    pass enums to callbacks
    remove deprecated cruft
  bookeeper more tightly integrated into zk?


Feel free to respond to this email with any thoughts, suggestions, etc...

Regards,

Patrick

Reply via email to