The way to do parallel scanning is with a map/reduce job and
TableInputFormat. This does all the work of parallelizing the scan,
as well as whatever work you were doing.
-Bryan
On May 10, 2008, at 1:49 PM, Daniel Leffel wrote:
Is there a parallel scanner (I didn't see it in the documents)? How
hard
would it be to create one that scans over regions on different servers
simultaneously? I mean, obviously, iteration order would not be
deterministic then, but that'd be ok. Would that actually make
table scans
faster?
