You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kylin.apache.org by hongbin ma <ma...@apache.org> on 2016/05/18 04:16:32 UTC

Re: expansion rate so high

I guess there is one or more dimensions with very high cardinality? have
you read http://kylin.apache.org/blog/2016/02/18/new-aggregation-group/?
there might be some inspirations there.

also, are the four fixed length dimensions all text? are there any
relationships between the dimensions? Why not using dict for them, are they
super-high cardinality columns?

On Mon, May 16, 2016 at 10:21 AM, Peng <pe...@outlook.com> wrote:

> Hi,
>
>    Is it normal when the expansion rate is about 1600% ?
>
>    My  cube :     about one hundred million data;
>
> fact table has one lookup table;
>
> 5 dimensions , in which 4 dimensions' encoding are fixed length,
> separately
> the length are 9, 8, 8, 33 ;
>
> 2 measures,
>
>  finally the expansion rate is about 1600.
>
>
>
> Thanks
>
> Peng
>
>


-- 
Regards,

*Bin Mahone | 马洪宾*
Apache Kylin: http://kylin.io
Github: https://github.com/binmahone

Re: expansion rate so high

Posted by Li Yang <li...@apache.org>.
Fixed length encoding is the least efficient encoding and a cause of the
high expansion. Consider use dictionary, or some custom encoding.

On Wed, May 18, 2016 at 12:16 PM, hongbin ma <ma...@apache.org> wrote:

> I guess there is one or more dimensions with very high cardinality? have
> you read http://kylin.apache.org/blog/2016/02/18/new-aggregation-group/?
> there might be some inspirations there.
>
> also, are the four fixed length dimensions all text? are there any
> relationships between the dimensions? Why not using dict for them, are they
> super-high cardinality columns?
>
> On Mon, May 16, 2016 at 10:21 AM, Peng <pe...@outlook.com> wrote:
>
> > Hi,
> >
> >    Is it normal when the expansion rate is about 1600% ?
> >
> >    My  cube :     about one hundred million data;
> >
> > fact table has one lookup table;
> >
> > 5 dimensions , in which 4 dimensions' encoding are fixed length,
> > separately
> > the length are 9, 8, 8, 33 ;
> >
> > 2 measures,
> >
> >  finally the expansion rate is about 1600.
> >
> >
> >
> > Thanks
> >
> > Peng
> >
> >
>
>
> --
> Regards,
>
> *Bin Mahone | 马洪宾*
> Apache Kylin: http://kylin.io
> Github: https://github.com/binmahone
>