Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: sar has the correct information or the vmstat? (student projects versus "official" software)

RE: sar has the correct information or the vmstat? (student projects versus "official" software)

From: Mark W. Farnham <mwf_at_rsiz.com>
Date: Wed, 30 Mar 2005 10:18:01 -0500
Message-ID: <KNEIIDHFLNJDHOOCFCDKKELKGDAA.mwf@rsiz.com>


Connor raises an interesting question: Does the original source of software being a student project correlate with superiority or inferiority of a tool as compared to software sourced "officially?"

This is a question of fact likely to go unanswered in a rigorous sense. Likely if the domain of competition is all student projects, then "official" will win, but I suspect student projects that survived into common use have the edge over all "official" software. MS-DOS, for example, was "official" software, while the Dartmouth Time Sharing System was largely a student project. A tiny number of people ever hacked DTSS.

As to sar and vmstat, I've gotten varied results over the years as to which more accurately portrayed various values on which releases of which operating systems. Unless you have an independent physical measure of an operational componment that both vmstat and sar measure or way to establish a known load and have both sar and vmstat report, you can't tell for sure. Even then some of the measurements will get "Hiesenberged." Finally, at the bottom line there is the frequency of flushing of kmem structures or their analog, so you just really can't get around the problems of time slice aggregations and averaging for some of the metrics being tracked by both these tools.

Usually you can find release notes or bug reports for a specific combination if one or the other is far enough off to be unreliable on a specific combination of hardware and operating system. In lieu of such notes, if you can reduce the problem to a particular system and release you can probably run a few load tests to decide whether one or the other is better in your situation.

Good luck!

mwf

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org]On Behalf Of Connor McDonald Sent: Tuesday, March 29, 2005 6:20 PM
To: oracle-l_at_freelists.org
Subject: Re: sar has the correct information or the vmstat?

I remember at a Sun presentation once the presenter remarked that "vmstat was the official tool
and 'sar' started life as a student project and should be treated as such".

I dunno if thats truth or fiction.

Cheers
Connor

--
http://www.freelists.org/webpage/oracle-l
Received on Wed Mar 30 2005 - 10:27:59 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US