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

Posted to user@hawq.apache.org by Yongzhu Li <yo...@gmail.com> on 2016/03/28 07:25:41 UTC

shared_buffers setting on hawq

Greetings,

Any one had good experience to share about adjusting shared_buffers on
HAWQ? We hope to get some performance improvement out of this but failed.


We followed the instruction on
http://hawq.docs.pivotal.io/docs-hawq/guc_config-shared_buffers.html.

But our code search shows shared_buffers is only used for heap table (which
is supported in greenplum but not HAWQ), and HAWQ only supports append only
table.

Any one, especially from the hawq dev community, could you comment or
confirm the statement above?

Thanks a lot.

Eric

Re: shared_buffers setting on hawq

Posted by Lei Chang <le...@apache.org>.

Hi YongZhu,

HAWQ only uses heap tables for catalog. And catalog is typically not the
bottleneck for olap queries. So you will not get performance improvement
when adjusting the value.

Cheers
Lei



On Mon, Mar 28, 2016 at 1:25 PM, Yongzhu Li <yo...@gmail.com> wrote:

> Greetings,
>
> Any one had good experience to share about adjusting shared_buffers on
> HAWQ? We hope to get some performance improvement out of this but failed.
>
>
> We followed the instruction on
> http://hawq.docs.pivotal.io/docs-hawq/guc_config-shared_buffers.html.
>
> But our code search shows shared_buffers is only used for heap table
> (which is supported in greenplum but not HAWQ), and HAWQ only supports
> append only table.
>
> Any one, especially from the hawq dev community, could you comment or
> confirm the statement above?
>
> Thanks a lot.
>
> Eric
>