[rrd-users] Re: Total num bytes graph solved (i think)
savage at savage.za.org
Sat Jul 19 21:27:41 MEST 2003
> > I know someone called Chris here who posted the subject *shrugs* had a
> > similar problem. I don't quite know how accurate my graph is (i'm not
> > confident with my mathematical skills :), but i think it match my
> > ok. I actually did as Chris suggested using PREV in a CDEF, but i
> > that since the first PREV is unknown the rest of the calculation fails,
> > just added a test and replaced PREV with 0 if PREV was unknown, as
> Sigh. How many times do I need to post this?
> The total amount of traffic equals the average amount of traffic
> multiplied by the total amount of time.
*sigh*, how many times do *we* still have to post this...
> Now GPRINT totalin, totalout and/or total.
We don't want to GPRINT, we want to GRAPH it!! (More precisely, it will be a
AREA, or a STACK that should be used - at least to answer our questions
A GPRINT will simply print at the bottom of the graph, something like total
traffic used: 1.65GB. Which, is useless because you cannot see WHEN the
1.65GB was utilised, only over what period (a total in other words). With a
AREA or a STACK, you can for example see each and every "interval" how your
*TOTAL* traffic (each interval) climbs higher and higher.
In other words, at 'rrdtool graph --start=whatever', whatever represents a
0byte total, and then over each period (interval), the total is incremented
and graphed on a AREA. Should the total exceed a predefined (in a CDEF)
threshold, for example 15GB, we change the color and STACK it on top of the
The end result, is that you have a GRAPH, showing for example in green,
total traffic utilised free of charge, plus on top of that in red, traffic
over 15GB that a client is charged for and must pay. The user viewing the
graph, can immediately see WHEN this "threshold" has been exceeded, and
depending on the slope of the graph (always going up), he can see how
quickly, or how much total data was transmitted or received. Should the
slope for example be around 10 degrees, the network was relatively idle for
a couple of hours or whatever. Should the slope for example be 60 degrees,
it's safe to presume that someone downloaded / uploaded a hell of allot of
data over those intervals...
Frankly, I don't know how to explain this any better....
I managed to get something to this extent working, but it doesn't work
properly because the total calculated always resets to zero when the
Interface resets (It counts AVARAGE as well as GAUGE in the same
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help mailto:rrd-users-request at list.ee.ethz.ch?subject=help
More information about the rrd-users