Using only for Luwak

Mark Phillips mark at basho.com
Tue Jan 17 13:22:32 EST 2012


Hi Nicholas,

As much as we would love you to use Riak for your application, at this
point you're advised to do it without Luwak. As few weeks ago we
informed the list that we won't be actively developing Luwak [1]
following the forthcoming Riak 1.1 release. While it was a useful
piece of code conceptually, architecturally it proved problematic to
support and extend.  Dizzy provided some details on the specifics a
few days back [2].

On the topic of Riak's feasibility without Luwak for your use case,
how large are the files you were planning to store in it?

Mark

[1] http://lists.basho.com/pipermail/riak-users_lists.basho.com/2011-December/006905.html
[2] http://lists.basho.com/pipermail/riak-users_lists.basho.com/2012-January/007170.html


On Tue, Jan 17, 2012 at 9:56 AM, Nicholas Young
<nicholas at nicholaswyoung.com> wrote:
> I'm building an app in which Riak will (at least ideally) serve as both a database and object/blob store.
>
> I considered running a single
> Riak cluster and installing Luwak, then dumping the large amounts of binary data alongside my JSON data, but the Luwak part of my cluster will grow much faster than the other sets of data. (I'm hoping to use Luwak as the storage engine behind a small CDN.)
>
> Two questions:
>
> Is there a way to somehow separate my Luwak data from my regular K/V data? If so, I would like to put higher performance disks in the machines hosting Luwak nodes.
>
> If there isn't a solution to Q 1, then maybe I can run two clusters: one for my K/V data, and the other for Luwak alone? If I chose to go this route, does running Riak for Luwak alone cause any problems I should be aware of?
>
> Thanks for your assistance.
>
> Nicholas
>
> _______________________________________________
> riak-users mailing list
> riak-users at lists.basho.com
> http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com



More information about the riak-users mailing list