Re: Release Note Changes - Mailing list pgsql-hackers

From Pavan Deolasee
Subject Re: Release Note Changes
Date
Msg-id 2e78013d0712102008w576dee81g7071e3c631bcef04@mail.gmail.com
Whole thread Raw
In response to Re: Release Note Changes  (Andrew Dunstan <andrew@dunslane.net>)
Responses Re: Release Note Changes  (Bruce Momjian <bruce@momjian.us>)
List pgsql-hackers
<br /><br /><div class="gmail_quote">On Dec 10, 2007 6:43 PM, Andrew Dunstan <<a
href="mailto:andrew@dunslane.net">andrew@dunslane.net</a>>wrote:<br /><blockquote class="gmail_quote"
style="border-left:1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="Wj3C7c"><br
/><br/>Pavan Deolasee wrote:<br />><br />><br />><br />><br />><br />> I don't agree completely. HOT
updatesis just one significant benefit of<br />> HOT and is constrained by the non-index column updates. But the
other<br />> major benefit of truncating the tuples to their line pointers applies to<br />> HOT as well as COLD
updatesand DELETEs. This should also have<br />> a non trivial positive impact on the performance.<br />><br />
>There might be few scenarios where HOT may not show any improvement<br />> such as CPU-bound applications, but I
amnot sure if its worth mentioning.<br />><br />><br /></div><br /><br />Um, I don't understand. I freely admit
thatI haven't kept up with all <br />the nuances of the HOT discussions, but this bit has totally eluded me,<br />so
pleaseelucidate.<br /><br /></blockquote></div><br /><br />One of the improvements of HOT is to truncate a DEAD tuple
toits<br />line pointer. A DEAD tuple could be an old version of an updated <br />tuple or a deleted tuple. When a
tupleis truncated, the space used<br />by the line pointer can not be reused (until the index entries are removed).<br
/>Butthe space used by the actual tuple can be reused for a later update, after <br />the page is defragmented. Note
thatthis defragmentation can happen<br />outside of a VACUUM.<br /> <br />This gives us an ability to run VACUUM less
frequentlyon a table. We<br />still need to run VACUUM to remove the line pointer bloat, but may be less <br
/>frequentlyfor the given percentage of bloat. IMHO this should have a positive<br />effect on performance atleast in
anIO bound scenario.<br /><br /><br />Thanks,<br />Pavan<br /><br />-- <br />Pavan Deolasee<br />EnterpriseDB     <a
href="http://www.enterprisedb.com">http://www.enterprisedb.com</a> 

pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Problem of a server gettext message.
Next
From: "Hiroshi Saito"
Date:
Subject: Re: Problem of a server gettext message.