Topic Options
#1316 - 01/07/05 06:02 PM Max Y Value of Graphs
Anonymous
Unregistered


My apologies if this has been already addressed elsewhere.

With the graph resolution display period set to any value greater than 60 minutes the auto scaling Y value (latency) is always some inaccurate, averaged-looking value rather than the true maximum Y value of the displayed time frame. Aside from the undesirable arbitrary discontinuity in behavior from the accurately displayed max Y value at resolutions of 60 minutes or less, this appears to simply be a radically inaccurate manner to designate graph scale. What's the story here?

Top
#1317 - 01/07/05 06:11 PM Re: Max Y Value of Graphs
Pete Ness Offline



Registered: 08/30/99
Posts: 1106
Loc: Boise, Idaho
You're not the first to ask this question. We have a knowledge base article on this:

http://www.nessoft.com/kb/20

If you have questions/concerns after reading that, please let us know.

Oh - here's a discussion on this:

http://www.nessoft.com/forums/showflat.php?Number=655

There are a couple of others as well, but this is a good one talking about reasons why it behaves like this.

- Pete


Edited by Pete Ness (01/07/05 06:32 PM)

Top
#1318 - 01/07/05 07:07 PM Re: Max Y Value of Graphs [Re: Pete Ness]
Anonymous
Unregistered


Thank you Pete. I see what you're doing, and appreciate the value of displaying a trend graph.

I would argue that such a graph be clearly identified as not portraying true scaled values, but rather a proprietary average trend function (Objects In The Rear View Mirror May Appear Closer Than They Are).

A casual, but nevertheless critical observer (e.g., your ISP) has no way of knowing that the maximum Y scale with a value of say, 120 ms in a submitted graph, refers to some processed/averaged proprietary value, and not to the peak latency of the time frame being displayed - which might be actually be 1020 ms.

In analogous scientific and financial charting software, the option to display similar propietary scales is also present, but never utilized without being routinely, fully and unambiguously identified as such. The Y scale value should never be a mystery, or a surprise, no matter how clever, or well intentioned it's design. How can you trust a graph as serious data documentation if you have to guess what the Y scale is?

From a purely utilitarian perspective, there are times when I'd like to look at a 12 hour stretch of graph and immediately see only the peak values and have an accurate scale so I can assess all their relative, true values - even if most plot points are lost in the noise below. Again, this is commonplace scientific and financial graphics functionality.

I suspect the above sounds a little severe - so let me say also that your software is easily the very best of its kind - thanks for all your hard work!

Bob

Top

Search

Who's Online
0 registered (), 18 Guests and 1 Spider online.
Key: Admin, Global Mod, Mod