[mrtg] Re: MRTG: value without division per time interval (no gauge)

[Pete Templin]

"Counters can be reported as an absolute value, or a rate of change per some interval."

In the real world, they can also be reported as an amount of change.  Ever had your PHB ask you for how many N you've done since the last time you did your X report?  The PHB probably wants to know N.  The PHB doesn't want to know that there have been 0.5 N per day or 15 N per month, they want to know that there have been 12 N since the last time (s)he told the CEO how many N have been done.  The PHB and CEO both know the time interval involved, but maybe they're just looking for a quantity.

In my case, I see 100-500 events per five minutes.  I don't want to see the default 0-2 events per second, because I lose tons of resolution.  If I do "perminute", I get 20-100 per minute, and I get more resolution, but not the same resolution as if I chart the actual difference.  If I do "perhour", I get 1200-6000 events per hour, but the number shown represents a five-minute sample.  That is much harder to explain to non-technical people.  It's much the same as trying to explain that you were going 4MPH during part of the commute and 26MPH during another part of your commute, but your commute only lasted 12 minutes and you only had to go 4.7 miles.

Perhaps it's just time for a "perfiveminute" option or a generic "perinterval" option.

Pete Templin
Senior Staff Engineer
TexLink Communications
(210) 892-4183
pete.templin at texlink.com


-----Original Message-----
From: Jerry Heidtke [mailto:jheidtke at fmlh.edu]
Sent: Friday, September 12, 2003 8:47 AM
To: Pete Templin; Alex van den Bogaerdt; MRTG MRTG
Subject: RE: [mrtg] Re: MRTG: value without division per time interval (no gauge)


Counters can be reported as an absolute value, or a rate of change per
some interval. Reporting a difference between sampled values is just a
special case of a rate of change per an interval, where the reporting
interval equals the sample interval.

By default, mrtg treats sampled values as counters, normalizes changes
to a 300 second interval (regardless of the actual sample interval), and
reports the derived rate of change per second. Each of these conditions
can be changed with available options.

It sounds like what you want is the "perminute" option. RTFM for more
details.

Jerry

-----Original Message-----
From: Pete Templin [mailto:pete.templin at texlink.com]
Sent: Friday, September 12, 2003 8:24 AM
To: Alex van den Bogaerdt; MRTG MRTG
Subject: [mrtg] Re: MRTG: value without division per time interval (no
gauge)


Yeah, MRTG works in rates.  But I don't want to know how many BGP
updates occurred per second, I want to know how many BGP updates
occurred since the last time I checked.  I don't want per minute, I
don't want per hour; they don't give a resolution that's extremely
useful.  I see 100 to 500 updates every five minutes.  That'd be 20-100
updates per minute, but if the graph is 135 pixels tall, I'd rather not
lose 4/5 of my precision.  I don't want to write an external program
interface when there's an OID readily available to me.  Is that so hard
to understand?

Besides, if you were to draw a matrix of the combinations possible
(calculate the difference or read the current value, divide by the
interval or not), you'd see that there are four combinations but only
three of them can be achieved in current form (native, gauge, absolute).
Is it really so hard for folks to realize that, and maybe consider that
there is a valid use for it? 

*putting fresh flame suit on*

Pete Templin
Senior Staff Engineer
TexLink Communications
(210) 892-4183
pete.templin at texlink.com


-----Original Message-----
From: Alex van den Bogaerdt [mailto:alex at ergens.op.het.net]
Sent: Friday, September 12, 2003 8:05 AM
To: MRTG MRTG
Subject: [mrtg] Re: MRTG: value without division per time interval (no
gauge)

On Fri, Sep 12, 2003 at 07:56:20AM -0500, Pete Templin wrote:

> What's needed is an option ("events"?) that calculates the difference
between the previous poll and the current poll, and simply displays
that.  Default calculates the difference and divides by the interval.
Gauge shows the current poll and doesn't divide by the interval.
Absolute shows the current poll divided by the interval.  "Newoption"
would calculate the difference but not divide by the interval.

No, this is not needed.  MRTG works in rates.  Numbers per second.
GAUGE is a way to tell MRTG that the number is already a rate.
It is still a rate, per second.

Events per <unknown time> are rates but they are not in the
events per second format.  *If* you write a front end (the external
program interface comes to mind) that translates "events per unknown"
into "other number of events per second" then you can use GAUGE just
fine.

Alex
--
http://www.googlism.com/index.htm?ism=alex+van+den+bogaerdt&type=1

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi


Confidentiality Notice: This e-mail message, including any attachments,
is for the sole use of the intended recipient(s) and may contain
confidential and privileged information.  Any unauthorized review, use,
disclosure or distribution is prohibited.  If you are not the intended
recipient, please contact the sender by reply e-mail and destroy all
copies of the original message.

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi