O2's Tu Go service No call volume on incoming or outgoing calls

[#1]

Local host is now turned off
I have also been power cycling my router to get the names of some affected and also some working servers.
pcl-ag07 - not working
pcl-ag04 - not working
ptn-bng01 - works just fine
pcl-ag05 - works just fine
To test each server I tried the three URL's above and also made a test call using Tu Go.  In each case a "works just fine" server loaded the affected page post haste and the Tu Go caller at the other end of the virtual blower could be heard just fine.  If I have stated that it is not working then a test call was connected but no inbound audio could be heard and the web pages would not load.
I've also got some other good news - my friend should have had her PN DSL turned on today.  It's not fibre but may add to our knowledge of this issue once I have done some testing - I am heading around at lunch time to get her new kit set up then perform some URL and Tu Go tests!
A test call using Tu Go with the local host (DMZ) disabled makes not difference.

[#2]

So I was having a nosey at where the servers are located, turns out that I used to work (in a disaster scenario) at Telehouse.  I had not heard of City Life Line House so I googled it.
Ironically (or maybe not as this could be a connected issue) but the http://www.city-lifeline.co.uk website does not load - on ANY of the servers.  All I get is "server not responding" error, the same error with the other URL's.
Please note that http://www.city-lifeline.co.uk is the ONLY URL that does not load at all on O2's 4G connection where as all other tested URL's and Tu Go work 100% of the time.

[#3]

Update on the 20-40 minute "working just dandy" window.
I am currently connected to gateway pcl-ag05, when first joining the rack everything worked fine.  Within 20 mins of connection I am again not able to connect a Tu Go call and the web pages in question also now fail to load.  Double checked the rack and I am still on the same one.
There was a Telehouse server that worked fine.  I am going to try to get back to one of their servers and see if the behaviour is the same.

[#4]

Can you eliminate the Airport and try the 582?

[#5]

I eliminated the Airport already, twice.
Once I tried my laptop via ethernet using the PN router, the issue persisted.
I then tried my laptop making a direct PPPoE connection to the fibre modern - again the issue persisted.
I am currently on ptw-ag01 - works just fine - Telehouse West, I have been connected for about 5 mins now and everything is working as it should do, Tu Go, web, PS3 and iMessage.
Am waiting on the 20-40 min failure window to see of the Telehouse servers also fail after this time period.

[#6]

I am still connected to gateway ptw-ag01, again after about 20 mins the connection reverted to an unusable state.
After checking the home page for both City Lifeline and also Telehouse I noticed that both of their home pages would not load.  To rule out that this is or isn't the same issue that is affecting me I opted to call both of the providers with one question; why their home page would not load.
City Lifeline responded by advising that their webmaster was updating their website and that is why it was down.  This did not quite ring true with me as I know that the BBC can update a web page and not have the whole site go down.  I also can update my own website and not have down time.  Whatever their actual issue is at least they spoke to me with dignity.
Telehouse however were not so respectful, the man that answered was dismissive of my question about their home page.  He stated that he did not have any daily requirement to access the home page and so how would he know that it was up.  My logical response was apt, I stated that if a company who's main business is to serve the internet services to the nation is not able to advise on the status of their own home page then what does that say about the businesses core services?  I asked to speak to someone who could answer my simple question.  At this point he placed the phone on hold (no hold button was pressed and I could hear the man speaking to another man in the background).  He was rude, called me a "c0¢k", his colleague said "tell the Cee U Next Tuesday to pi$$ off" along with several other expletives.  Having worked with Telehouse I am pleased to say that this is not a "one off" issue and that in the past I have had to deal with incidences like this.  The difference back then was that I was a direct customer of theirs and could raise a formal complaint directly with them.
This time around I am a member of the public that took the liberty of recording the foul language used and I am happy to report that I will be posting the conversation online for the public to dissect.
I've spoken to PN support about this, they have advised to raise a complaint with a link to the copy of the recording of the call so that you can investigate.  I am still one of their costumers by proxy and I will not be spoken to like that.
In an hour or so I will head around to my friends house, once I have set up her DLS I will perform the same tests there and let you know the results of the test.

[#7]

Sorry, I should have added the City Lifeline site doesn't work for me also. http://www.telehouse.net/ works fine.
What a strange problem you've got. Is there any kind of log you can get from the Apple Airport?
Your TBB BQM looks fine - so I am struggling to see where the problem might be.

[#8]

It is indeed a very strange set of problems.  The web is digital, 0's and 1's.  It usually works or it doesn't.  I have rarely seen services that work fine apart from one feature (inbound audio on Tu Go as an example).
I do not know of any way to get the logs from the newer AirPort's.  You used to be able to do that with the older models/older Airport Admin Tool but not with these new ones that I have got now.  I don't think that is the issue anyway.  Am about to head to my friends house.  I think that I will take my Airport over to hers and test it there.  We have already ruled it out as a point of failure though so I am expecting it to work just fine.

[#9]

Are you running anything on your network that could be opening lots of ports on your router?  (P2P application?)

[#10]

Yes I do P2P at times (not all the time), the PS3 also opens multiple ports but again it's not on all of the time.
However we have ruled this out as a point of failure by using the laptop (that doesn't run P2P) to create PPPoE direct connection and I've also tested the connection using just my laptop and the plusnet router.  The issue was seen in both of these "environments" that had been stripped down to just a single user session with nothing running but a web browser.

[#11]

Sorry!  You did point that out.    (Serves me right for just dipping into the thread)  Really strange.
Will chat to Dave and see if we can come up with some other ideas

[#12]

When you used your laptop for a PPPoE connection, did it have the same symptoms that sites work for 20-40 mins and then stop?
It's a strange one because the BQM looks fine and your traceroutes looked ok (although I was able to reach the destination for two of them).
Can you run through your network setup? Are you specifying your own DNS servers?

[#13]

Thanks for thinking outside the box for me.
I have just repeated the test with the laptop and fibre with ethernet cable for the PPPoE connection.  I took the liberty (in the name of keeping it simple) of creating a new "test" login for my laptop.  This ensures that there is no 3rd party software running on the laptop whilst performing the test.  Doing this is almost as good as getting a brand new machine and turning it on for the first time.
It took 5 reboots of the line to pick up ptw-ag03 before the pages in question would load.  As soon as I had a "good" connection confined I started the timer then started refreshing the page every few seconds.  2 minutes and 45 seconds into the test the connection went "bad".
I do not use 3rd party DNS servers, I have (in the name of testing) used Google's - this has not resulted in any change of behaviour.  I would be happy to discuss my home network setup at length if you feel that it would help.  This being said however I am not happy about discussing detailed specs of my home network on a public forum.  This aside I think that it is conclusive that this issue in no way lies within my home network environment.  My home network is not your average run of the mill setup, I run a server, home automation, media streamers and Freesat to name a few.  I understand it's a complex beast which is why, early on, I suggested that we try a paired down version of a single user set up so that we could rule my home network out as the point of failure.

[#14]

This is a bit of a mystery.
Can you pin-point when it happened?
Are you using a Mac? It might be worth installing nmap to see if you can ping those sites that don't work on TCP Port 80.

[#15]

It first came to my attention way back last year.  I have several calls to support where I ask why are pages loading slowly and querying my line speed.  Line tests were done but no fault was found.
The timescales rule out the recent change to the firewall settings and also suggest that this issue was apparent from the start.  However it was only when I realised that I had a 1 in 6 (approx) chance of power cycling my modem to get a (albeit short lived) good connection.  It was at this point that all of my seemingly unrelated issues could be brought under the one issue.
Yes I do have a Windows free home, nmap has been duly installed;
Starting Nmap 6.46 ( http://nmap.org ) at 2014-04-22 18:09 BST
Note: Host seems down. If it is really up, but blocking our ping probes, try -Pn
Nmap done: 1 IP address (0 hosts up) scanned in 3.20 seconds
MacBook-Pro:~ ken$ sudo nmap -Pn www.nyc-architecture.com
Starting Nmap 6.46 ( http://nmap.org ) at 2014-04-22 18:09 BST
Nmap scan report for www.nyc-architecture.com (66.96.132.67)
Host is up (0.031s latency).
rDNS record for 66.96.132.67: 67.132.96.66.static.eigbox.net
Not shown: 997 filtered ports
PORT    STATE SERVICE
21/tcp  open  ftp
554/tcp  open  rtsp
7070/tcp open  realserver
Nmap done: 1 IP address (1 host up) scanned in 12.80 seconds