Re: Decent VACUUM (was: Buglist) - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Decent VACUUM (was: Buglist)
Date
Msg-id 22619.1062045596@sss.pgh.pa.us
Whole thread Raw
In response to Re: Decent VACUUM (was: Buglist)  (paul_tuckfield@yahoo.com (Grant Succeeded))
List pgsql-hackers
paul_tuckfield@yahoo.com (Grant Succeeded) writes:
> The best for me by far, is to get the OS to *not* cache stuff.  As
> long as the database uses the information it inherently has available,
> it can make far more effective use of the same amount of memory the OS
> would have used to cache the whole filesystem.

This is a very long-running debate in this community, and I think at
this point the majority opinion contradicts yours.  The OS buffer cache
has a number of fundamental advantages over what the database can do,
the most obvious being that it actually knows how much free memory is
available for buffer cache at any instant.  Also, the existing DBs that
take the we'll-do-the-buffer-management approach are designed around
thirty-year-old OS technology.  I'm not inclined to accept this position
without some substantial evidence in its favor.
        regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Possible bug in update?
Next
From: Larry Rosenman
Date:
Subject: Re: Beta2 Tag'd and Bundled ...