Hi Vijay, Chukwa+Hbase is definitely in the realm of possibility. I had been playing with Hbase as one of the candidate for replacing mysql in Chukwa. An abstract class with load(fs,directory) method should be a good start point for processDemuxOutput to load data to multiple pipelines (Hbase, mysql, or pig). It's fairly trivial to clean up the dataSource hashmap. DataSource hashmap should be moved into the DataLoader to decide what to do with the sequence file(s).
Don't worry, if you don't plan to write the driver for HICC. I am actually interested to write a reference implementation to read data from Hbase, and render on HICC for the fun of it. Regards, Eric On 5/31/09 9:13 PM, "Vijay" <[email protected]> wrote: > Hi, > > i would like to contribute hbase adapter for chukwa.... i am writing it > extending the classes of postprocess as they are written for Mysql can we > just separate it to make it and both Mysql and Hbase extend the abstract > class or an interface? i will submit the hbase code once i have a working > set of it.... i am not sure about hicc as we are planning to use our own > display with flex, and already have a custom parser working. > > Regards, > </VJ>
