Buckets and riak instances
timothy at getintheloop.eu
Mon Oct 26 09:36:55 EDT 2009
I agree - it would be less natural.
Perhaps its just me being used to RDBMS but im wondering about the
best strategy of keeping data of a riak cluster separate between
buckets. My first thought is that I would rather now have lots of Riak
nodes running on different ports as that means a lot more
administration. For instance, if I was hosting several clients data in
one Riak cluster, its nice to know that one clients data could not be
in anyway connected to the others (you know how customers worry about
Thanks for tolerating my questions
On 26 Oct 2009, at 12:57, Bryan Fink wrote:
> On Mon, Oct 26, 2009 at 5:05 AM, Timothy Perrett
> <timothy at getintheloop.eu> wrote:
>> Thanks for your response - most helpful. One thing that is not
>> clear on the
>> NYC nosql slides ( http://riak.basho.com/nyc-nosql/ ) on slide
>> number 21 is
>> weather or not Artist, Album and Track are all in the same bucket?
>> Im guessing they are, but its kinda tough to know :-) In this
>> example, is
>> the bucket also called "artist"?
> Hi, Tim. In slide 21, Artists, Albums, and Tracks are definitely each
> in their own bucket. The reason they must be in their own bucket has
> to do with the link-traversing syntax. In either the Erlang or URL
> syntax, the first element of the link query is a "bucket match",
> meaning "follow all links that point to objects in bucket X".
> So, the example says:
> 1. Start at the "artist" keyed "REM"
> 2. Follow all links that artist has to objects in the "album" bucket
> 3. Follow all links those albums have to objects in the "track"
> This simple example could be transformed into having all objects in
> one bucket by using the link tag to differentiate among types instead
> of the bucket name, but that seemed less natural in this case.
> riak-users mailing list
> riak-users at lists.basho.com
More information about the riak-users