Is Riak suitable for s small-record write-intensive billion-records application?

Sean Cribbs sean at basho.com
Mon Oct 22 09:55:47 EDT 2012


On Mon, Oct 22, 2012 at 9:41 AM, Joshua Muzaaya <joshmuza at gmail.com> wrote:

>
> This is not meant to down-market Riak, but you mentioned Billions of
> records. Riak storage is known to have a few issues as data grows to
> billions. However, Couchbase 2.0 has been kinda battle tested, using SQLite
> at the storage layer.


I'm not sure how you are backing up this assertion (citation?). We have
customers storing terabytes of data with billions of records (using
LevelDB). SQLite as the storage engine does not necessarily make Couchbase
more stable or battle-tested, just better known among the wider community.
On the other hand, we have spent the last eighteen months, in collaboration
with the original authors at Google, hardening and improving the
performance of LevelDB to support large deployments, and it now performs
very well (although we still can do better).

I'm not saying that the OP should not consider Couchbase, but either
product is going to require more than just a casual knowledge of ops to
deploy, manage and maintain. Clustered datastores are not for the
faint-of-heart, so if the OP can do it in a local SQL database, then he
should.

Sean Cribbs <sean at basho.com>
Software Engineer
Basho Technologies, Inc.
http://basho.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20121022/05f6ad99/attachment.html>


More information about the riak-users mailing list