Secondary Index Query across multiple buckets
rusty at basho.com
Thu Dec 15 11:41:58 EST 2011
Yes, in this scenario, a single query under a large bucket will be faster,
for all of the reasons that you mentioned... connection setup/teardown,
query processing overhead, and seeking to/reading index information from
On Thu, Dec 15, 2011 at 11:07 AM, Jeremy Raymond <jeraymond at gmail.com>wrote:
> Thanks Rusty. What would perform better, running multiple queries across
> several buckets or a single query under one large one? I'd guess the later
> because of whatever overhead is involved in setting up the connection to
> Riak and whatever Riak does to setup and run a mapred/index query.
> On Thu, Dec 15, 2011 at 10:59 AM, Rusty Klophaus <rusty at basho.com> wrote:
>> Hi Jeremy,
>> Currently, Secondary Indexes are scoped per bucket. There is no good way
>> to run a Secondary Index query across multiple buckets, short of running
>> multiple queries.
>> You are correct in that you could store the items in a single bucket
>> provided that they all have unique keys. In the LevelDB backend (and most
>> backends), buckets are a logical separation, not a physical separation, so
>> either approach will use the same amount of resources. (The Innostore
>> backend is the exception to this rule, it splits buckets into separate
>> On Thu, Dec 15, 2011 at 8:29 AM, Jeremy Raymond <jeraymond at gmail.com>wrote:
>>> Let me pose the question a different way. What is a good way to pull out
>>> objects across multiple buckets based upon some secondary data that is not
>>> their key?
>>> The items in the buckets all have unique keys. Maybe move them all into
>>> a single bucket (millions of items) and then I could use secondary indexes
>>> to pull the find them?
>>> If the keys of my objects are unique does organizing them into buckets
>>> really gain me anything over just putting them all in a single bucket and
>>> using secondary indexes to find them?
>>> On Sat, Dec 10, 2011 at 2:05 PM, Jeremy Raymond <jeraymond at gmail.com>wrote:
>>>> Is it possible to run an index query across multiple buckets at once?
>>>> Something similar to how you can feed a map query with bucket/key pairs,
>>>> I'm looking to run an index query with bucket/index pairs.
>>> riak-users mailing list
>>> riak-users at lists.basho.com
>> Rusty Klophaus (@rustyio)
>> *Basho Technologies, Inc.*
Rusty Klophaus (@rustyio)
*Basho Technologies, Inc.*
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the riak-users