You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@hbase.apache.org by Michael Dagaev <mi...@gmail.com> on 2009/02/17 14:44:30 UTC

Row lock overhead

Hi, all

     Currently,  I use a batch update (~100 rows per request).

If I want to use row locks now I will have to throw away the batch
and update each row separately. Is it correct ? Does it add much
more overhead ?

Looks like it would be better to use optimistic locking in this case.
Is it correct?

Thank you for your cooperation,
M.

Re: Row lock overhead

Posted by Andrew Purtell <ap...@apache.org>.

Or you could try transactional tables, contributed by Clint
Morgan, available in 0.19.

http://hadoop.apache.org/hbase/docs/current/api/org/apache/hadoop/hbase/client/transactional/package-summary.html

Certainly this will add some overhead, but if opportunistic
concurrency will work for your application, you can try it out.

  - Andy

> From: Michael Dagaev
> If I want to use row locks now I will have to throw away
> the batch and update each row separately. Is it correct ?
> Does it add much more overhead ?