Re: proposal: be smarter about i/o patterns in index scan - Mailing list pgsql-hackers

From Sailesh Krishnamurthy
Subject Re: proposal: be smarter about i/o patterns in index scan
Date
Msg-id mjqy8nqgyo1.fsf@cs.berkeley.edu
Whole thread Raw
In response to proposal: be smarter about i/o patterns in index scan  ("Jeffrey W. Baker" <jwbaker@acm.org>)
List pgsql-hackers
Yes, fetching a RID list from an index scan, sorting 'em and then
fetching from the table would be a very appreciable speedup for many
queries. I would imagine that all the commercial engines do this (db2
calls it a sorted RID-list-fetch) .. and this has in fact been
discussed on -hackers before.

One issue for this is that there could be a slip between the cup and
the lip .. ie., between the fetch from the index, the sort, and the
fetch from the table a vaccuum could have taken place rendering TIDs
invalid. This should however not be a show-stopper .. surely all
that's needed is a lock against vaccuuming in the presence of a
tid-list-fetch. 

-- 
Pip-pip
Sailesh
http://www.cs.berkeley.edu/~sailesh




pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: Call for 7.5 feature completion
Next
From: Neil Conway
Date:
Subject: FYI: away for the summer