You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@hbase.apache.org by Ted Yu <yu...@gmail.com> on 2010/12/29 23:01:49 UTC

bloom filter types

In 0.90,
    /**
     * Bloom enabled with Table row as Key
     */
    ROW,
    /**
     * Bloom enabled with Table row & column (family+qualifier) as Key
     */
    ROWCOL

Is there wiki / doc on which type to use in various scenarios ?

Thanks

Re: bloom filter types

Posted by Stack <st...@duboce.net>.

Here is link to the 0.90.0 BF doc Ted:
http://people.apache.org/~stack/hbase-0.90.0-candidate-2/docs/blooms.html
 Its from some doc Nicolas wrote way back.

@N Yeah, if you want to add a bit to the book or elsewhere (We can
link to the latter).

St.Ack

On Wed, Dec 29, 2010 at 2:06 PM, Nicolas Spiegelberg
<ns...@fb.com> wrote:
> I don't think there's an explicit wiki.  Which option depends on whether
> your use case is calling get() for entire rows or for specific columns in
> a row. It also depends on analyzing your workload to determine how likely
> a row will be in every store file vs. a specific column.  Also, since a
> row is a coarser granularity than a column, it might be good to switch to
> a row bloom if your BF starts taking up too much space.  I guess this
> sounds like a nice article for me...
>
> On 12/29/10 2:01 PM, "Ted Yu" <yu...@gmail.com> wrote:
>
>>In 0.90,
>>    /**
>>     * Bloom enabled with Table row as Key
>>     */
>>    ROW,
>>    /**
>>     * Bloom enabled with Table row & column (family+qualifier) as Key
>>     */
>>    ROWCOL
>>
>>Is there wiki / doc on which type to use in various scenarios ?
>>
>>Thanks
>
>

Re: bloom filter types

Posted by Nicolas Spiegelberg <ns...@fb.com>.

I don't think there's an explicit wiki.  Which option depends on whether
your use case is calling get() for entire rows or for specific columns in
a row. It also depends on analyzing your workload to determine how likely
a row will be in every store file vs. a specific column.  Also, since a
row is a coarser granularity than a column, it might be good to switch to
a row bloom if your BF starts taking up too much space.  I guess this
sounds like a nice article for me...

On 12/29/10 2:01 PM, "Ted Yu" <yu...@gmail.com> wrote:

>In 0.90,
>    /**
>     * Bloom enabled with Table row as Key
>     */
>    ROW,
>    /**
>     * Bloom enabled with Table row & column (family+qualifier) as Key
>     */
>    ROWCOL
>
>Is there wiki / doc on which type to use in various scenarios ?
>
>Thanks