On 10/30/07 7:11 PM, "edward yoon" <[EMAIL PROTECTED]> wrote:
> Well then, What do you think about hbase storage option for Pig?
I think that is a grand idea. Pig could provide high throughput batch
analysis of hbase databases.
> If you're positive about hbase storage option,
> Pig will manipulate the hbase databases and the ultimate goal of hbase shell
> and pig could be the same (as for SQL-related discussions, we should leave it
> on the side.)
The ultimate goals are clearly relatively similar. I also shouldn't be
discouraging you from trying to do anything you would find interesting.
I just think that it would slow down experimentation on both sides (hbase
and pig) if they were tied too closely together at an early stage.
Hbase shell is, from what I can see, quite a hot bed of experimentation with
syntax and interface. Pig is all about how to express map-reduce
computations easily and efficiently. In the short run, those cultures
should not be forced to live by each others assumptions.
