Read-Write-Read for enforcing read-repair

Les Mikesell lesmikesell at gmail.com
Mon Aug 6 15:59:26 EDT 2012


On Mon, Aug 6, 2012 at 2:33 PM, Ersin Er <ersin.er at gmail.com> wrote:
>
> I am new to Riak and its concepts and just trying to see whether it can fit
> into my domain of applications. I am working on core Java server software
> systems (no enterprise, no heavy sql) and I am very used to having control
> on all of data in memory with strict semantics.
>
> So for one of the cases I am evaluating Riak, I need to make sure that when
> a transaction (or request) is executed and finished, db value it resulted
> also has been persisted at its final form which does not have any conflicts
> waiting to be resolved by other processes. Result of the transaction is sent
> back to the client and it's unacceptable for this result to be invalidated
> by another process which overwrites this result in db due to conflict
> resolution routine.
>
> The question is whether a read-write-read sequence is acceptable or applied
> pattern for such cases or are there any other solutions to this problem? Or
> am I just expecting too much?

I don't think riak has any transaction-like concepts or any way to
give one writer precedence over another in the case of concurrent
updates.

-- 
   Les Mikesell
     lesmikesell at gmail.com




More information about the riak-users mailing list