Re: 2.1.103 cache size still too big

Paul M. Hirsch (hirsch@instrumental.com)
Fri, 5 Jun 1998 12:17:19 -0500 (CDT)


On Fri, 5 Jun 1998, Riley Williams wrote:

> Hi Edgar.
>
> >> We should have a maximum percentage of memory that can be
> >> used for the caching of _one_ file, possibly tunable by
> >> counting the number of pagefaults for this file vs. the
> >> total number of pagefaults in the system.
>
> > One day you'll have so much *tunables* that you have to
> > swap /proc/sys :-(
>
> > IMO, memory managment is going the wrong way...
>
> Personally, I'd prefer memory management to be one of those things
> that just gets done RELIABLY, without having to worry about tuning
> some collection of variables that are meaningless to most of the
> people using the system. In my book, OPTIMAL is of considerable less
> importance than RELIABLE - that's why I use Linux in preference to
> WinDoze in the first place...

I would agree that, out of the box, memory management parameters should
be set to reliable levels. However, in the area of memory management
(as well as some other kernel parameters), one size does not fit all.
Tunable parametrs are of interest and benefit to myself, and others, and
there is no need to reduce their number or eliminate them. They are a
great feature of Linux, and I would like to see even more put in. (Not
TOO many. Another, more important feature of Linux is "Lack of Bloat".)

That's my 20% of a dime.

-Paul

------------------------------------------------------------
Paul M. Hirsch
System Administrator/Support Engineer
Instrumental, Inc.
vox-> 612.603.0295 fax-> 612.602.0781
smtp-> hirsch@instrumental.com http-> www.instrumental.com
------------------------------------------------------------

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu