Kitz Forum

Broadband Related => Router Monitoring Software => Topic started by: geep on June 08, 2013, 10:02:27 PM

Title: "Fixing my internet with R and Python"
Post by: geep on June 08, 2013, 10:02:27 PM
This has got to be interesting to somebody...
http://www.r-bloggers.com/fixing-my-internet-with-r-and-python/

Edit: Some of the plots at the above link are just wrong!
Better to look at the original version of the document:
http://jacobsimmering.com/2013/02/21/UsingPythonAndRToTroubleshootConnectivityProblems.html


Cheers,
Peter
Title: Re: "Fixing my internet with R and Python"
Post by: burakkucat on June 08, 2013, 11:23:30 PM
 :hmm:  My four word summary of the original blog post is Interleaved versus Fast Path.
Title: Re: "Fixing my internet with R and Python"
Post by: kitz on June 16, 2013, 12:10:21 PM
Thanks geep, interesting re the python script.

Mega busy over the past few and next few weeks so Id not read this, but just having had a quick look at his post Im not sure that what he is saying is correct about Interleaving

Quote
I could open a terminal and run ping google.com or some other site and it would come back with very reasonable times. Some of the time (2-3% of all packets), however, it would throw huge pings on the order of 300 to 700 ms.

Interleaving does NOT cause variable ping times.
Nor is it something that would occur at the ISP network*

Depending upon the depth of Interleaving (http://www.kitz.co.uk/adsl/interleaving.htm) (& error correction) then it would add a fixed amount of latency (ie say 20ms) to each & every hop outside his LAN on to the internet.   Nor would it ever through latency up 300 > 700 ms.

I dont have time to analyse his graphs, but from what he says he would seem to suggest that the greatest variance is occurring not at the router, nor the first hop (when interleaving would be applied) some variance that the 3rd (ISP) hop but largely on the remote hop.

The fact that the latency is so variable and what he says at the 'ISP network', then that to me would imply congestion...  possibly either on the ISP gateway or the ISP <-> CoLo hop to the ISP transit (http://www.kitz.co.uk/adsl/equip3.htm) provider.

The whole idea of interleaving is to add robustness during the transit of data on the local loop - not at the ISP network.   Switching to FAST mode does not suddenly lop 700ms off the latency!!!!! nor stop the huge variance of latency.

/me thinks that something else was going on..  and perhaps a switch of ISP gateway during the change from Interleaving to Fast mode is more likely to have been what cured the problem ;)

Title: Re: "Fixing my internet with R and Python"
Post by: kitz on June 16, 2013, 12:12:32 PM
PS

If anyone wants or needs such a script but arent up to coding it themselves in Python, perhaps try Ping Plotter (http://wiki.kitz.co.uk/index.php?title=Network_tools#Ping_Plotter) or WinMTR (http://wiki.kitz.co.uk/index.php?title=Network_tools#WinMTR)

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwiki.kitz.co.uk%2Fimages%2F9%2F9f%2FPingplotter.gif&hash=65715c9689109d83fd0921e4482a4633ebfd4b33)
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwiki.kitz.co.uk%2Fimages%2Fd%2Fdb%2FWinMTR.jpg&hash=67178b0341a8d4aa3f11bfc2801acf3831a37659)