[smokeping-users] Standard Deviation in Smokeping

[Rui Meireles]

Hi. I've been using Smokeping for a while. Please correct me if I'm wrong.

 

Smokeping shows a value for the Standard Deviation (SD) on the bottom of the
graphs, labeled SD.

I believe the values Smokeping shows for sd is the "SD of the AVERAGES" and
not the "AVERAGE of the SDs", as it should be.

I'll explain in this example. Imagine 2 graphs:

 

Graph #1:

*	all the medians are around 20ms, and there's hardly any smoke.

                                                               i.      STD
in each pool would be very small, "average of the SDs" would be VERY SMALL.

Graph #2:

*	all the medians are around 20ms, but there is lots of smoke.

                                                               i.      STD
in each pool would be very big, "average of the SDs" would be VERY BIG.

 

But the value shown in Smokeping graphs is the "SD of the AVERAGES", and it
will be VERY SMALL IN BOTH GRAPHS, because the medians are all close to
20ms!

 

I would like a value for SD that could tell me information about the "amount
of smoke" in the graph, but I believe this would only be possible it there
was a datasource (ds) for SD in each RRD, and I don't think this is how
things are being done.

 

 

I use this command to obtain an AVERAGE RTT:

rrdtool graph CR-BT1.rrd -s -2592000 DEF:median=CR-BT1.rrd:median:AVERAGE
PRINT:median:AVERAGE:%lf |tail -1

 

*	How can I get the "SD of the AVERAGEs"?
*	How can I get the "AVERAGE of the SDs"? (If it's possible. If not,
how can I obtain a value that reflects the amount of smokein a graph?)

 

 

I'd appreciate a lot any help you could give me!

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/smokeping-users/attachments/20090121/9aa0e1f8/attachment.html