From: | Jim Nasby <Jim(dot)Nasby(at)BlueTreble(dot)com> |
---|---|
To: | konstantin knizhnik <k(dot)knizhnik(at)postgrespro(dot)ru> |
Cc: | Anastasia Lubennikova <a(dot)lubennikova(at)postgrespro(dot)ru>, <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Batch update of indexes |
Date: | 2016-02-05 00:21:51 |
Message-ID: | 56B3EB1F.4010805@BlueTreble.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 2/4/16 1:37 AM, konstantin knizhnik wrote:
>> >My suspicion is that it would be useful to pre-order the new data before trying to apply it to the indexes.
> Sorry, but ALTER INDEX is expected to work for all indexes, not only B-Tree, and for them sorting may not be possible...
> But for B-Tree presorting inserted data should certainly increase performance.
> I will think about it.
I wasn't talking about ALTER INDEX.
My theory is that if you're doing a large DML operation it might be more
efficient to update an index as a single bulk operation, instead of
doing it for each tuple.
If you want to do that, then you need an efficient method for finding
everything that a DML statement changed. That's the exact same thing we
need to support statement-level triggers being able to reference NEW and
OLD. It's probably also what we need to support incremental update matviews.
If we had such a capability then we could add options to the AM
infrastructure to allow indexes to support doing bulk maintenance as
well as per-tuple maintenance (or even support only bulk maintenance...)
I don't think any of that has anything to do with ALTER INDEX.
--
Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX
Experts in Analytics, Data Architecture and PostgreSQL
Data in Trouble? Get it in Treble! http://BlueTreble.com
From | Date | Subject | |
---|---|---|---|
Next Message | Andres Freund | 2016-02-05 00:31:59 | Re: "using previous checkpoint record at" maybe not the greatest idea? |
Previous Message | Jim Nasby | 2016-02-05 00:16:30 | Re: Way to check whether a particular block is on the shared_buffer? |