Data stored in Search *and* KV ?

Rusty Klophaus rusty at basho.com
Thu Nov 4 14:21:35 EDT 2010


Hi Nicolas,

Search stores a representation of the indexed object so that it knows what
to do later if you delete the object, change the object, or re-index the
object with a different analyzer. The representation allows us to delete the
old inverted index entries correctly.

We have considered adding a toggle to disable storage of the representation
object for write-once data, but it's not something the system currently
supports.

Best,
Rusty

On Thu, Nov 4, 2010 at 10:33 AM, Nicolas Fouché <nicolas at silentale.com>wrote:

> Hi,
>
> Following the discussion with seancribbs and Tv on #riak:
> http://pastie.org/private/acxekqyapbk7fz1hcp8udg
>
> I only index documents with the bucket precommit hook and I do all my
> searches via Map/Reduce queries.
> I'm not sure how Search works, if there is an inverted index and a
> forward index. I'm just "worried" about storing full documents in the
> Search forward index *and* in KV, as I would just need the Search
> forward index to store the document id (=key in KV).
>
> Do you consider this as a problem ? Is it how the way Search works,
> and I would have to live with data stored twice ?
>
> -- Nicolas
>
> _______________________________________________
> riak-users mailing list
> riak-users at lists.basho.com
> http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20101104/052b2569/attachment.html>


More information about the riak-users mailing list