Re: Performance degradation of REFRESH MATERIALIZED VIEW

Hi,

Based on the investigation and (lack of) progress so far, I'll revert
part of the COPY FREEZE improvements shortly. I'll keep the initial
7db0cd2145 changes, tweaking heap_multi_insert, and remove most of
39b66a91bd (except for the heap_xlog_multi_insert bit).

This should address the small 5% regression in refresh matview. I have
no other ideas how to fix that, short of adding a user-level option to
REFRESH MATERIALIZED VIEW command so that the users can opt out/in.

Attached is the revert patch - I'll get it committed in the next day or
two, once the tests complete (running with CCA so it takes time).

regards

On 5/25/21 12:30 AM, Tomas Vondra wrote:
> On 5/24/21 8:21 PM, Andres Freund wrote:
>> Hi,
>>
>> On 2021-05-24 12:37:18 +0200, Tomas Vondra wrote:
>>> Another option might be changes in the binary layout - 5% change is well
>>> within the range that could be attributed to this, but it feels very
>>> hand-wavy and more like an excuse than real analysis.
>>
>> I don't think 5% is likely to be explained by binary layout unless you
>> look for an explicitly adverse layout.
>>
>
> Yeah, true. But I'm out of ideas what might be causing the regression
> and how to fix it :-(
>
>>
>>> Hmmm, thanks for reminding us that patch. Why did we reject that approach in
>>> favor of the current one?
>>
>> Don't know about others, but I think it's way too fragile.
>>
>
> Is it really that fragile? Any particular risks you have in mind? Maybe
> we could protect against that somehow ... Anyway, that change would
> certainly be for PG15.
>
>
> regards
>

--
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company