In the presentation at Hadoop World NYC, "real time map-reduce" was
mentioned in the HBase presentation.

I guess that means to run user scripts/functions in HBase region server
process space (Just like BigTable can run user filters and sawzall)
Is there any more details on that? It might be hard to support arbitrary
map-reduce programs, so I guess there will be some restrictions?

-- 
Yours,
Zheng

Reply via email to