Message-ID: <37814c0413f9d36338cca990a90180eba47f74c1.camel@jotel.com.pl>
Subject: Index merge join
From: Marek Mosiewicz <marek.mosiewicz@jotel.com.pl>
To: pgsql-hackers@postgresql.org
Date: Sat, 22 Apr 2023 17:15:35 +0200
Content-Type: multipart/alternative; boundary="=-LeZ2l9tPhhR2d80CezUF"
User-Agent: Evolution 3.48.0 (3.48.0-1.fc38) 
MIME-Version: 1.0
Archived-At: <https://www.postgresql.org/message-id/37814c0413f9d36338cca990a90180eba47f74c1.camel%40jotel.com.pl>
Precedence: bulk

--=-LeZ2l9tPhhR2d80CezUF
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi,

Best way to restrict query on many columns is to have index on these
columns.
BUT with faster and faster IOPS is it not already feasable to make
merge
join on indexes ? Then it could be possible to have just index on
single columns
for more rare queries. Index which have column and oid/primary key.
Then make merge join on all columns restreicted/sorted.
Same for full text search (word,relevance,id)

Best regards,
=C2=A0 =C2=A0Maek Mosiewicz

--=-LeZ2l9tPhhR2d80CezUF
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: quoted-printable

<html><head></head><body><div>Hi,</div><div><br></div><div>Best way to rest=
rict query on many columns is to have index on these columns.</div><div>BUT=
 with faster and faster IOPS is it not already feasable to make merge</div>=
<div>join on indexes ? Then it could be possible to have just index on sing=
le columns</div><div>for more rare queries. Index which have column and oid=
/primary key.</div><div>Then make merge join on all columns restreicted/sor=
ted.</div><div>Same for full text search (word,relevance,id)</div><div><br>=
</div><div>Best regards,</div><div>&nbsp; &nbsp;Maek Mosiewicz</div><div><s=
pan></span></div></body></html>

--=-LeZ2l9tPhhR2d80CezUF--