Yokozuna max practical bucket limit

Elias Levy fearsome.lucidity at gmail.com
Mon Apr 22 17:11:26 EDT 2013


On Mon, Apr 22, 2013 at 1:21 PM, Ryan Zezeski <rzezeski at basho.com> wrote:

>
> For Solr this would be a delete-by-query which isn't nearly as expensive.
>

How so?  My understanding is that Lucene's index looks pretty much like a
LSM-tree, and that any delete (instead of an index drop), will result in
the writing a of a tombstone into a new segment, which eventually will have
to merged.  So a delete is as costly as a write plus the attendant costs of
merging.

Even if that is solved, Riak Search will have other tradeoffs such
> as substantially reduced feature support compared to Yokozuna as well as
> reduced performance for many types of queries.  But I do agree many indexes
> (thus cores) could pose a problem for Yokozuna.
>

Agreed.  I think Yokozuna is a great step forward and I am looking forward
to it.  I just wanted to point out this scaling limitation that results
from the every index on every node architecture.

Elias Levy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20130422/6f341b00/attachment.html>


More information about the riak-users mailing list