Re: GIST optimization to limit calls to operator on sub nodes - Mailing list pgsql-performance

From Pujol Mathieu
Subject Re: GIST optimization to limit calls to operator on sub nodes
Date
Msg-id 53B11450.2010001@realfusio.com
Whole thread Raw
In response to Re: GIST optimization to limit calls to operator on sub nodes  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: GIST optimization to limit calls to operator on sub nodes  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-performance
Le 29/06/2014 22:30, Tom Lane a écrit :
> Emre Hasegeli <emre@hasegeli.com> writes:
>> Pujol Mathieu <mathieu.pujol@realfusio.com>:
>>> I made my own index to handle specific data and operators. It works
>>> pretty fine but I wonder if it was possible to optimize it.
>>> When I run my operator on a GIST node (in the method
>>> gist_range_consistent) it returns "NotConsistent" /
>>> "MaybeConsistent" / "FullyConsistent".
>>> NotConsistent -> means that all subnodes could be ignored,
>>> gist_range_consistent return false
>>> MaybeConsistent -> means that at least one subnode/leaf will be
>>> consistent, gist_range_consistent return true
>>> FullyConsistent -> means that all subnodes/leaves will be
>>> consistent, gist_range_consistent return true
>>>
>>> So like with the "recheck flag" I would like to know if there is a
>>> way to notify postgres that it is not necessary to rerun my operator
>>> on subnodes, to speedup the search.
>> I do not think it is possible at the moment.  The GiST framework can
>> be extended to support this use case.  I am not sure about the
>> speedup.  Most of the consistent functions do not seem very expensive
>> compared to other operations of the GiST framework.  I would be
>> happy to test it, if you would implement.
> I don't actually understand what's being requested here that the
> NotConsistent case doesn't already cover.
>
>             regards, tom lane
>
>
Hi,
The NotConsistent case is correctly covered, the sub nodes are not
tested because I know that no child could pass the consistent_test.
The MaybeConsistent case is also correctly covered, all sub nodes are
tested because I don't know which sub nodes will pass the consistent_test.
My problem is with the FullyConsistent, because when I test a node I can
know that all it's childs nodes and leaves will pass the test, so I want
to notify GIST framework that it can't skip consistent test on those
nodes. Like we can notify it when testing a leaf that it could skip
consistent test on the row. Maybe I miss something on the API to do
that. On my tests, the "recheck_flag" works only for leaves.
Thanks
Mathieu



pgsql-performance by date:

Previous
From: Niels Kristian Schjødt
Date:
Subject: Re: Guidelines on best indexing strategy for varying searches on 20+ columns
Next
From: Craig Ringer
Date:
Subject: Volatility - docs vs behaviour?