Re: Re: A separate table level option to control compression

Hi,

On Thu, Mar 21, 2019 at 3:10 AM Shaun Thomas <shaun(dot)thomas(at)2ndquadrant(dot)com>
wrote:

>
> I can't really speak for the discussion related to `storage.sgml`, but
> I based my investigation on the existing patch to `create_table.sgml`.
> About the only thing I would suggest there is to possibly tweak the
> wording.
>
> * "The compress_tuple_target ... " for example should probably read
> "The toast_tuple_target parameter ...".
> * "the (blocksize - header)" can drop "the".
> * "If the value is set to a value" redundant wording should be rephrased;
> "If the specified value is greater than toast_tuple_target, then
> we will substitute the current setting of toast_tuple_target instead."
> would work.
>

Thanks Shaun. Attached patch makes these adjustments.

> * I'd recommend a short discussion on what negative consequences can be
> expected by playing with this value. As an example in my tests, setting
> it very high may result in extremely sparse pages that could have an
> adverse impact on HOT updates.
>

Setting compress_tuple_target to a higher value won't be negative because
the toast_tuple_target is used for compression anyways when
compress_tuple_target is higher than toast_tuple_target. May be some
discussion in the paragraph related to toast_tuple_target can be added to
explain the negative impact of the high value.

I added a small discussion about negative effects of setting
compress_tuple_target lower though, per your suggestion.

Also added some details in storage.sgml as recommended by Sawada-san. Hope
this helps.

Thanks,
Pavan

--
Pavan Deolasee http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services