Automating gateway hoping

[#1]

Seems that for me there are 2 classes of gateway; one on which I get ~10ms pings, and one where it is closer to 15ms.  I was thinking of automating the process I do manually at the moment (ie check the ping latency after connection, and disconnect / reconnect in an attempt to get a faster gateway).  Can anyone think of a reason not to do this?  For the moment I was only going to do it at connection time, not constantly monitoring.

[#2]

pings to where? Do you get an extra 5ms latency to any destination, or just whatever you're pinging?

[#3]

I get an extra ~5ms latency for pretty much anything I expect, just ping is the easiest thing to see it on.  You can see the difference in the inbound ping on the graph in my sig - I disconnected earlier today and landed up on a worse gateway.  Once I realised, I disconnected and reconnected and came back onto a good one (took me a while to realise).
if I ping out to anywhere I see the same kind of increase, and my plan was to define a "good" ping time and on connect test the connection.  I've seen this phenomenon ever since joining plusnet.
Even 15ms ping is really good, so don't get me wrong I'm not complaining, but if for a little effort I can be more likely to have a good connection then I'll do it

[#4]

Do you game?

[#5]

Not unless you class being a meddler as gaming!

[#6]

You mean a fiddler    If it ain't broke don't fix it - not

[#7]

well probably a bit of both! These things intrigue me.  In other news, I'm currently building an OpenWRT image with pppd patched up to support proper ethernet MTU of 1500 over PPPoE, using baby jumbo frames on my WAN interface (you currently end up with 1492 MTU after the PPPoE overhead is taken into account).  I think this will lead to an increase in throughput...  It will at least occupy my fidget fingers for a while.

[#8]

Spotted any pattern in the latency increases?  Is it certain gateways?

[#9]

I'll pay more attention.  I know it is always seems to be a constant state for the "worse" gateways - once I'm connected to one it won't change til I connect to something else.  The last time I took note I had a slow response on pcl-gw01.  I've had good results on both pcl-gw02 (current) and ptn-gw01.
Today:
My Broadband Ping
19th:
My Broadband Ping
18th:
My Broadband Ping
9th:
My Broadband Ping
8th:
My Broadband Ping
7th:
My Broadband Ping
3rd:
My Broadband Ping

[#10]

I think you probably mean pcl-ag01 & pcl-ag02  & ptn-ag1 jimbof  
Would you like a list Kelly I've been compiling one, every time I change Gateways. I can certainly tell you which ones are rubbish (for me anyway)
To be fair I have noted some improvements, presumably because of some of the work you've been doing. I'll have tocompare my list with what jimbof has posted.

[#11]

Like a dummy I'm reading the wrong line in the traceroute...!
Good:
pcl-ag08
ptn-ag01
pcl-ag02

"Bad":
ptw-ag01
pcl-ag06  ?

[#12]

Now for me ptw-ag01 is quite good! I'll have to check for others alter, my list is on another computer.

[#13]

Well I hacked together a script last night and ran it for a couple of hours; basically it would connect up, work out what gateway it was on, then ping thinkbroadband's ping machine for a couple of minutes and display the average ping.  It then disconnected, reconnected, waited a minute and tested again.  The results as below.
For me last night over this number of tests it seems like all of pcl-ag01-06 could become quite slow ping wise, with pcl-ag03 putting in the worst performance of all.  Though those gateways aren't permanently bad it seems - If you look further down the list you can see often a gateway which has performed poorly can then have a much better result within a short timeframe, whereas my experience is that typically once connected with slow pings you are stuck at that ping for the session duration.  Weird.
18.115 pcl-ag03 - Sun May 26 00:25:04 BST 2013
14.036 pcl-ag01 - Sun May 26 00:15:51 BST 2013
13.647 pcl-ag06 - Sun May 26 00:00:29 BST 2013
13.591 pcl-ag02 - Sat May 25 23:57:24 BST 2013
13.426 pcl-ag01 - Sun May 26 00:34:17 BST 2013
13.315 pcl-ag05 - Sun May 26 00:43:30 BST 2013
13.297 pcl-ag04 - Sun May 26 00:37:21 BST 2013
12.097 pcl-ag06 - Sat May 25 23:23:37 BST 2013
10.244 ptw-ag04 - Sat May 25 23:35:54 BST 2013
10.179 pcl-ag04 - Sat May 25 23:29:45 BST 2013
10.062 pcl-ag01 - Sat May 25 23:26:41 BST 2013
10.029 pcl-ag04 - Sun May 26 00:18:55 BST 2013
9.914 ptw-ag02 - Sat May 25 23:51:16 BST 2013
9.898 pcl-ag01 - Sun May 26 00:06:38 BST 2013
9.893 pcl-ag06 - Sun May 26 00:21:59 BST 2013
9.836 ptw-ag02 - Sat May 25 23:48:11 BST 2013
9.826 ptn-ag02 - Sat May 25 23:45:07 BST 2013
9.807 ptn-ag02 - Sat May 25 23:38:58 BST 2013
9.732 ptn-ag04 - Sat May 25 23:32:50 BST 2013
9.661 ptw-ag03 - Sat May 25 23:42:03 BST 2013
9.511 pcl-ag06 - Sun May 26 00:09:42 BST 2013
9.480 ptn-ag02 - Sun May 26 00:03:33 BST 2013
9.470 ptn-ag01 - Sat May 25 23:54:20 BST 2013
9.420 pcl-ag03 - Sun May 26 00:12:46 BST 2013
9.391 pcl-ag04 - Sun May 26 00:28:08 BST 2013
9.352 pcl-ag04 - Sun May 26 00:31:12 BST 2013
9.290 ptw-ag04 - Sun May 26 00:40:26 BST 2013

[#14]

A couple of things you could try doing - especially as I would expect some variability at peak times -

• testing just after midnight might not give reliable results as a number of users will have scheduled large downloads because of the free overnight usage - not everyone is on unlimited usage remember.


• Stay on each Gateway somewhat longer and see how much variability you get over time - especially the ones that appear worse. What I find is that for me, slower gateways never get better than a minimum value, but they do get a lot worse at times, as do the better ones at times. I see you have a ping plotter running.

I've been plotting my pings lately to investigate a completely different issue and have seen some interesting patterns that I've yet to analyse and look at in depth - and to be sure it's nothing to do with the program I'm currently plotting with.

• When you connect to a Gateway, do a tracert to the same IP address each time (maybe the one you are pinging) and log it. See if any of the intermediate hops change for the same gateway and whether this shows any particular routing as being bad.


• Check your Router logs each time you connect to a Gateway and see if you are connecting via the same bras node each time.

If I think of anything else to suggest I'll post it.

[#15]

i would like to Add , for some reason any PCL gateways are bad for gaming , i always try to avoid them when i hop .
i have no idea why but it is between 1700 - 2100 they are at their worst, sorry i haven't any test to show this