[rrd-users] Re: Newbie RRD User - Questions

Alex van den Bogaerdt alex at slot.hollandcasino.nl
Tue Jul 24 00:04:31 MEST 2001


Schroeder, Dennis wrote:

> rrdtool create file.rrd --step 60 DS:Load:GAUGE:120:0:255
>  RRA:MAX:0.5:1:720        (RRA #1)

720 rows of one minute per PDP and one PDP per CDP.  This is a total
of 720*(1 minute step)*(1 steps per CDP) is 720 minutes is 12 hours.
Check.

[snip rest, also OK]

> #2. If I GRAPH "AVERAGE", does it automatically select the RRA based on the
> time period of the graph?

Yes, it will make sure to take one of the AVERAGE RRAs.  If possible,
it will be an exact match; else it will be the RRA that best matches
your request.  Note that "best" may be something different from what
you expect.  It is generally best to set the end time to an exact
boundary (not "now" but rather "`date +%H:00`" for instance) and to
have an RRA which will provide one CDP for each pixel (and thus a
time span on your graph which matches the RRA).  So, if you are
graphing 400 pixels per image, choose to graph 400 minutes ending
at "date +%H:%M", or 400 hours ending at "date +%H:00", or 400 days
ending at "date -u -d 00:00 +%s".  

If you rather like to graph a weeks worth of data, which is exactly
7*24*60=10080 minutes, choose to graph 10000 minutes on 400 pixels
with each pixel being 25 PDPs per CDP.  Have an RRA available that
has 25 PDPs per CDP and set the end time on a compatible boundary.
(again assuming a 400-pixel wide graph).

> #3. In a test I've been running, the AVERAGE and MAX graphs are almost
> exactly the same over short periods of time (hours), but if I graph a week
> or more the peaks seem to start to show up...is this common?

It is fairly simple.  If max(a,b,c...) is the function to find the largest
of the parameters, and avg(a,b,c...) calculates the average, try to find
a difference between "max(10)" and "avg(10)".  There is none.  Only once
the number of parameters increase (such as: "max(10,12)" vs. "avg(10,12)")
there will be a difference.  For the RRA with 1440 PDPs per CDP there's
much more chance of a larger value inbetween several smaller ones and
therefore the difference between the max and the avg will be much higher.

If you're graphing 800 minutes on a graph with 400 pixels, you
will be using 2 CDPs per pixel.  Depending on the function selected
(AVERAGE or MAX) you will do avg(a,b) or max(a,b) for each pixel.
However, avg(10,12) and max(10,12) differ only by one.  The next
pair might be avg(14,16) vs. max(14,16) which differs also by one.

For a graph of 1600 minutes and thus 4 minutes per pixel, you are
doing avg(10,12,14,16) vs. max(10,12,14,16) which has a difference
of 3.  Of course the numbers here are only examples and will differ
in reality.

HTH
-- 
   __________________________________________________________________
 / alex at slot.hollandcasino.nl                  alex at ergens.op.het.net \
| work                                                         private |
| My employer is capable of speaking therefore I speak only for myself |
+----------------------------------------------------------------------+
| Technical questions sent directly to me will be nuked. Use the list. | 
+----------------------------------------------------------------------+
| http://faq.mrtg.org/                                                 |
| http://rrdtool.eu.org  --> tutorial                                  |
+----------------------------------------------------------------------+

--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi



More information about the rrd-users mailing list