In a message written on Mon, Jul 08, 2002 at 06:22:00AM -1000, Internet Guy wrote:

Each one of these graphs shows abnormalities in the flow of internet data, such as "pits", spikes, square wave function graphs, clipping on some waveforms, etc.

You have drawn an incorrect conclusion. While several of these graphs have interesting shapes, they are not all abnormalities. Several are graphing errors. Several are human errors (graphing the wrong thing). Several are oddball traffic patterns due to strange customer usage. None of that indicates abnormal flow. One of my favorite things was working with people who do simulcasts of things like Victoria Secret's and The Drew Carey Show. At the top of the hour a huge spike, and at the bottom of the hour a perfect drop off. A half hour square function on every graph in the network. Doesn't mean anything was wrong. Rebalancing peers, dealing with outages, customers doing backups, installing and testing, etc. It all happens daily on a big network, and can produce some really odd graphs.

As there is no way to really verify the QUALITY or INTEGRITY of the data being displayed, then I submit as fact, that what is being shown here is really in a grey area, at best.

So who really knows how correct, the data which is being displayed on the MFN home page is really is ?

As I work for MFN, all I can say in a public forum is that relying on any one tool is foolish. If you use more than one tool, you can compare them and determine which is correct. Assuming that the graphs we make public are the only graphs we have would be a bad mistake. There's also the obvious, if you can "show int" on the box, you can always double check a graph, in real time anyway. -- Leo Bicknell - bicknell@ufp.org - CCIE 3440 PGP keys at http://www.ufp.org/~bicknell/ Read TMBG List - tmbg-list-request@tmbg.org, www.tmbg.org

On Mon, Jul 08, 2002 at 06:22:00AM -1000, Internet Guy wrote:

If you go to the MFN homepage & click on the graphs listed below, then you might see that possibly the data being displayed is both inaccurate, as well as misleading.

If you mean inaccurate as in "not updated in years" or "missing 95% of their circuits", than yes. If you mean inaccurate as in intentionally fudged, than no. Look at the bottom of http://west-boot.mfnx.net/traffic/ This article written by <noc@above.net> Fri Sep 3 17:07:26 PDT 1999 Copyright 1999 AboveNet Communications, Inc. And it wasn't even accurate then. They've obviously updated SOME things since then (the OC48/OC192s), but most of the circuits listed there have been gone for years, or havn't been graphed in years.

Go to SJC OC3 Los Angeles, to OC192 SJC3 to SJC4, to OC12 MaeW ATM, OC48 # 2 for IAD to NYR, IAD # 2 to PAIX VA OC48, DCA2 to DFW2 OC48, PAIX OC12 to Core1.sjc, NPA - DS3 to San Jose, LGA1 OC192#2 to IAD, LGA1 OC48 to Chicago, NYC Backbone OC192 to LGA2, NYC Backbone OC48 # 2 to core3.lga1, ETC...

Each one of these graphs shows abnormalities in the flow of internet data, such as "pits", spikes, square wave function graphs, clipping on some waveforms, etc.

Thats just how the internet works. Traffic is busting constantly. There is already a smoothing effect just because these are 5 minute samples. If you were to look at instantaneous values, they would be significantly more spikey. In general there are two well-known MRTG display problems: * When MRTG does not collect data, it displays a flatline for both in and out based on the last rate it calculated. This can be potentially misleading, and newer graphing systems like RRDTool simply leave the area blank. * Rateup breaks at > a signed 32 bit int of Bits/sec, 2^31Bps. If you find some graphs doing > 2Gbps, you'll see the lines mysterious drop off the graph completely. As for some of the truely bizaare graphs you cite, such as: http://west-boot.mfnx.net/traffic/sjc/sjc-lax-oc3.html I can't explain how that managed to get polled, but you'll note that the dates on all of those graphs are extremely old, such as June 15 2001. Lets just chalk it up to the first category of "noone cares enough to update those stupid MRTG's".

It is not easy to see HOW BAD the problem is with these Sundry data collection systems, UNTIL you expand the MRTG graph. Once this is done, then you can really see how bad the integrity of the collected data really is. A small MRTG graph really masks the problems associated with the data which is being displayed. With a larger graph, you definately see the problems associated with todays Sundry systems.

This is easily one of the stupidest things I have read in a long time. If you want to expand the graph, you must expand the frequency of data collection. There is nothing sundry about it. Put the crack pipe down and step away from the keyboard.

One would also wonder, that if this data collection system is used by MFN to generate bills for customers of MFN who are charged by the Megabyte, what these customers bills look like & HOW accurate these bills really are...

The graphs on west-boot have absolutily no relation to customer billing. They are barely updated by anyone, and only looked at by nosey people such as myself. MFN has better things to do with their limited resources right now, and if I was a customer I'd rather they take care of the important things first. -- Richard A Steenbergen <ras@e-gerbil.net> http://www.e-gerbil.net/ras PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 14 36 FE B6)