Questions around Riak "these days"

Mark Phillips mark at
Thu Aug 30 15:56:56 EDT 2012

On Tue, Aug 28, 2012 at 6:46 PM, Mike Oxford <moxford at> wrote:
> On Mon, Aug 27, 2012 at 10:56 PM, Mark Phillips <mark at> wrote:
>> Hi mox,
>> Welcome back. :)
> Thanks, good to be back.  Been an odd run lately.
>> As it stands right now, no. Still plain old boring Riak. If you want
>> custom resolution logic you'll still be doing that client-side with
>> something like Statebox or Knockbox.
> Gotcha, thanks.
>> As I said there's nothing in Riak now that solves this (though, to be
>> fair, it wasn't designed to act in this way). That said, all is not
>> lost. Sean Cribbs and Russell Brown are giving a talk called "Data
>> Structures in Riak" [0] at RICON in October. Joseph Blomstedt is also
>> giving a talk about some of the approaches that he and others on the
>> Basho team have discussed and may take to add consistency semantics on
>> top of Riak. (These talk details will be live tomorrow.) Now, there is
>> still a chance that, come October, nothing has changed and you'll
>> still be writing your own resolution logic and cursing Riak. But there
>> might be some conflict-free replicated light at the end of the
>> tunnel...
>> Perhaps you'll join us [1] at RICON to find out? :)
> I'm hoping, though not 100% sure at this time.  I was more curious along the
> lines of something like Statebox being merged into Riak directly.  You could
> at least write the resolver "into Riak" and instead of spawning siblings it
> could simply do the resolution right there, or keep track of which KVs have
> siblings and slow resolve them (or even do lazy-style on next retrieve.)
> Right now, if you have an Erlang client and a POST client you'll end up
> needing to write two identical resolvers.
> Then some "default" resolvers could be written, such as "append only."  No
> client work would need to be done at all, it would greatly extend the
> functionality, lower barriers to adoption into write-heavy environments,
> etc.
> Just a thought.   There are times for full custom-resolvers so you obviously
> cannot cover everything, but having the resolution done in Riak instead of
> spread out to the edges at the client-side might be of value.
> Fingers crossed for RICON...

Here are the full details on Joe's "Consistency" talk...

Hope to see you there...


> -mox

More information about the riak-users mailing list