Re: How to keep a table in memory?

In an attempt to throw the authorities off his trail, jd(at)commandprompt(dot)com ("Joshua D. Drake") transmitted:
> Andrew Dunstan wrote:
>>> I give this a +/- 1. Yes extremely heavy websites can do this
>>> *but* they require extremely expensive hardware to do so.
>>>
>> I expect extremely heavy websites to require extremely expensive
>> equipment regardless of the software they use. Cost was not the
>> issue raised by the OP.
>
> Cost is always an issue, even if implicit. If the person is so hung
> up on the idea of pushing things into ram there is a pretty good
> possibility they have priced out the 50 and 100 spindle devices
> needed to get the same type of performance.

I dunno; I had a chat about cacheing strategies today where it became
clear to me that when we migrate to 8.3, we'll need to re-examine
things because there has been *so* much change since some of our
present policy was created back in the 7.2 days.

(Pointedly, one of the reasons to want a separate cache DB was to cut
down on XID consumption by read-only processes, and that reason
evaporates in 8.3.)

I have seen enough naive analyses done that I wouldn't be inclined to
assume much of anything.

People can get mighty self-assured about things that they have heard,
whether those things have validity or not. Few things can get as
badly wrong as bad assumptions made about performance...
--
(reverse (concatenate 'string "moc.liamg" "@" "enworbbc"))
http://linuxdatabases.info/info/x.html
"When we write programs that "learn", it turns out that we do and they
don't." -- Alan Perlis