gateway problems 23/4 pm

[#1]

something not right here  

I did a gateway hop at about 9am this morning to get off ptw-bng01. My router & ONT showing green lights throughout, and a few other graphs showing the same issues. Did another hop (to ptw-ag04) about 17:45 and all seems well now

[#2]

I was about to start a new thread - I'm told our Internet connection had been "on and off" all day. It appeared to be totally stalled so I gateway hopped and ended up on ptn-bng01. Also switched to OpenDNS servers.
Not sure if this is totally conclusive but I think the bandwidth usage graphs seem a bit lower than usual, although it's hard to tell with only a 24-hour graph or a weekly graph.

[#3]

I think ptw-ag02 is suffering suddenly.  BOth of you were on it and the graphs aren't reporting right.  We're looking at it.

[#4]

I second the problems at 4pm onwards. The internet was connected, just stopped doing anything. Took a disconnect and reconnect to get back online but it did it 2 times. Finally pulled plug on the modem and it's done the trick and back online for now.

[#5]

@ejs - thank you for the pointer - interesting graphs
@Kelly - and what do the chinks tell us?
Midnight - ptn-ag1 - did it 'ditch' all of its sessions?
3am - pcl-ag02 similar
What about the chinks in some of the green graphs around 3pm?

Cheers,
Kevin

[#6]

i had the same    around 4.15 web went sloooooooow then disconected .
had to power off router to regain connection

[#7]

Quote from: Kelly

I think ptw-ag02 is suffering suddenly

heh - out of the frying pan  springs to mind   Only hopped off ptw-bng01 to try to avoid the 20:00-23.30 problems

[#8]

Had the same issue here, had to reboot the Netgear router to get back online.

[#9]

Not sure what gateway I was one. But similar issue.

[#10]

I thought something was wrong yesterday evening.
Mine was on / off continually from about 17:15 to 18:45. My fear was that 21CN would pick this up as a line issue and reduce my speed to a crawl overnight when does its checks

[#11]

During last night's peak time around 10pm i noticed that my throughput in both directions was lower than it normally is, there was also around 5ms of jitter, after some gateway hopping i eventually found a gateway that  provided the full throughput rates and had very next to no jitter, I have a feeling it was PTW 02, But by 02:00am it dropped my session,
IMO some of the time  it maybe the BTW endpoint that you connect to  responsible for poor throughput, from what i understand  these change every time you connect/swap gateway, Do these end point also change  even if you are continually connected for a long time?, i ask because this would also explain why  you can hop to gw that gives you  good performance for a week, but then it's jittery and throughput drops at peak times ,

[#12]

There's a definite problem with HTTP latency. It is very noticeable.
Always difficult to demonstrate conclusively due to the number of variables, so I can't be bothered doing extensive testing (or arguing about it,) but the attached may help to give an indication.
It is the first 10 packets of two separate captures of just the 44K of HTML being downloaded from
www.theregister.co.uk/2008/09/30/reg_ts_and_cs/
I picked that target because it's a fast site and a simple page. Not a scientific choice.
You can see that the TCP latency reflects the ICMP latency in the ping below. All good (apart from a bit of wavering). But, looking at packet number 6, there is considerable extra delay at the start of the HTTP transfer in the second capture. I did a few: these are the maximum and minimum, but there were various values in between.

Pinging www.theregister.co.uk [92.52.96.89] with 32 bytes of data:
Reply from 92.52.96.89: bytes=32 time=5ms TTL=54
Reply from 92.52.96.89: bytes=32 time=5ms TTL=54
Reply from 92.52.96.89: bytes=32 time=4ms TTL=54
Reply from 92.52.96.89: bytes=32 time=4ms TTL=54
Reply from 92.52.96.89: bytes=32 time=5ms TTL=54
Reply from 92.52.96.89: bytes=32 time=4ms TTL=54
Reply from 92.52.96.89: bytes=32 time=4ms TTL=54
Reply from 92.52.96.89: bytes=32 time=5ms TTL=54
Reply from 92.52.96.89: bytes=32 time=4ms TTL=54
Reply from 92.52.96.89: bytes=32 time=4ms TTL=54
Ping statistics for 92.52.96.89:
    Packets: Sent = 10, Received = 10, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 4ms, Maximum =  5ms, Average =  4ms

It could, of course, be the web server, or something to do with BT, or unrelated. This alone isn't enough to draw any conclusion from, but, given that www surfing performance seems to be much worse than it has been across a range of sites, I'm putting it down to instability in PlusNet's traffic management systems. I'm not seeing any packet loss, and throughput seems OK.
I don't know whether it's related to particular gateways or not. I'm on PCL-AG07 and have no plans to spend time hopping around.
This may be useful extra info; or it may not. I offer it "as is".