[rrd-developers] Feature suggestion - OP_TVAL

Jeremy Hinton jgh at visi.net
Mon Mar 4 18:07:07 MET 2002



	On working with several monitoring tools, i've noticed a area that
RRD based tools seem to be lacking: the ability to visualize a clear
correlation between curent and historical data on a single graph. IE, show
me (on one graph) todays data vs yesterdays data for a single data source.
I was thinking about how this would be done, and a CDEF operator seems
like a decent way, something like: <vname>,<time>,TVAL. When processed
this would return the value of <vname> at <time>. IE:

DEF:input=router.rrd:ds0:AVERAGE \
CDEF:oldinput=input,LTIME,86400,-,TVAL \
AREA:oldinput#00cc00:Yesterdays Avg \
LINE1:input#0000ff:Curent input

	One complication is that if <time> far enough back, you'd have to
pull the data from a different RRA. The CF to use would be identified by
the CF defined in the DEF for <vname>. Since (unlike MRTG) rrd puts
current data in to all RRDs, you wouldnt have to worry about missing data
from overlapping RRAs. With this ability, you could make some fairly
complex graphs, using CDEFs to graph the current value against say a
computed avergage of the last 7 days. Something like this would be very
useful in doing quick visual trend analysis. Yes, the granularity of the
data would be lost as the "time window" slid further back, but even a
comparison to a lower resolution historical data set could be quite
useful. I know you can do some of this now with multiple graphs, but it
would much easier (and quicker) to notice trend differences on a single
graph. 

	I took a quick look at rrd_graph.c, and how the OP_PREV and
OP_VARIABLE sections, as this construct would seem to be something of a
combination of those. However, i think something like this (if possible) 
is beyong my immediate ability (leastways without a week of trying to
familiarize myself with the code inner-workings and internal data
structures). My hope is that this message might inspire someone out there
to believe what a great feauture this is ;). If not, well, hopefully i'll
get around to taking a whack at it one of these days. 

- jeremy

// Jeremy Hinton		                  	VisiNet
// jgh at visi.net				    http://www.visi.net
// Network Operations		               Newport News, VA
// A little nonsense now and then is relished by the wisest men 



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



More information about the rrd-developers mailing list