It would be great to have sequential scan of index in Postgres like we have Index Fast Full Scan in Oracle.
Oracle Index FFS reads whole index and only index in the same way as full table scan (by several sequential blocks
(definedby parameter) )
STEPAN YANKEVYCH
Software Engineering Team Leader
Software Engineering Manager
OCA
Office: +380 322 424 642 x 58840 Cell: +380 96 915 9551 Email: stepan_yankevych@epam.com
Lviv, Ukraine epam.com
CONFIDENTIALITY CAUTION AND DISCLAIMER
This message is intended only for the use of the individual(s) or entity(ies) to which it is addressed and contains
informationthat is legally privileged and confidential. If you are not the intended recipient, or the person
responsiblefor delivering the message to the intended recipient, you are hereby notified that any dissemination,
distributionor copying of this communication is strictly prohibited. All unintended recipients are obliged to delete
thismessage and destroy any printed copies.
-----Original Message-----
From: Dmitry Dolgov <9erthalion6@gmail.com>
Sent: Wednesday, October 30, 2019 17:15
To: Stepan Yankevych <Stepan_Yankevych@epam.com>
Cc: stepya@ukr.net; pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #16089: Index only scan does not happen but expected
> On Wed, Oct 30, 2019 at 04:10:06PM +0100, Dmitry Dolgov wrote:
> > On Wed, Oct 30, 2019 at 02:18:09PM +0000, Stepan Yankevych wrote:
> >
> > >> And in fact at least in my tests this was indeed faster
> > Can you show execution plan when index is used?
> > Does it uses Index only scan or index scan and table scan?
>
> Yes, at the bottom of the plan there was an Index Only Scan going
> through l1_snapshot_201811 (although while populating the table I've
> set values only for not nullable columns).
To clarify, Index Only Scan was there only after changing random_page_cost/seq_page_cost. As I mentioned in the
previousemail, original plan was also parallel seq scan.