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

Posted to user@phoenix.apache.org by 孙文隆 <we...@icloud.com> on 2017/08/29 14:23:16 UTC

How to specify the point lookup chunks number(parallelism) at query

Hi there,

I am optimizing an sql that is compiled to point lookups. Since sometimes there could be many thousands of "in" conditions in the sql, I would like to increase the parallelism number for better performance. I know that I can increase the salted bucket number or decrease the guide post width to increase the parallelism number. But is there any way to specify the parallelism at the runtime of JDBC so that I can change it according to the number of "in" conditions?


Best regards,
Ryan

Re: How to specify the point lookup chunks number(parallelism) at query

Posted by 孙文隆 <we...@icloud.com>.

Thanks James,

It is a good idea. The situation is that my customer will give different numbers of "in" conditions randomly. So sometimes there is 10 point lookups for a sql while sometimes there could be 100 thousand point lookups. If I change the guidepost width to a smaller digit
, the parallelism number would be too many for the 10-point-lookup SQLs.

Best regards,
Ryan



发自我的 iPhone
> 在 2017年8月29日，下午10:43，James Taylor <ja...@apache.org> 写道：
> 
> granularity

Re: How to specify the point lookup chunks number(parallelism) at query

Posted by James Taylor <ja...@apache.org>.

The finest granularity we have available is to specify a guidepost width at
the table level with the GUIDEPOSTS_WIDTH=## property. This will take
effect for that table next time stats are collected (on a major compaction
or when collected manually through the UPDATE STATISTICS command).

On Tue, Aug 29, 2017 at 7:23 AM 孙文隆 <we...@icloud.com> wrote:

> Hi there,
>
> I am optimizing an sql that is compiled to point lookups. Since sometimes
> there could be many thousands of "in" conditions in the sql, I would like
> to increase the parallelism number for better performance. I know that I
> can increase the salted bucket number or decrease the guide post width to
> increase the parallelism number. But is there any way to specify the
> parallelism at the runtime of JDBC so that I can change it according to the
> number of "in" conditions?
>
>
> Best regards,
> Ryan