Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Alex Atkin UK]

Having the test box does give you some clout though, as if it clocks in slow enough it could drag their average down and they wont want to start advertising as being slower than everyone else.

[skyeci]

hmm now I have started seeing something a bit odd unless this is known about but I guess I will have to speak with zen about it as its not been a problem until a few days ago. Zen told me I am on a btw circuit and following a random reboot of my opnsense box a few days ago  I notice my latency shot up to almost double what it would normally be.. It now seems when ipv6 is enabled and I end up on lo0-0.bng4.wh-man.zen.net.uk I get around 14ms yet when I disable ipv6 and I seem to go via vt1.cor1.lond2.ptn.zen.net.uk using just ipv4 my latency goes back to 6/7ms..

Dont wish to hi-jack the thread but this just seems odd and has never been an issue before.

any thoughts?, why is the man gateway giving me a much greater latency?

IPV6 enabled

     Server: Swish Fibre - London (id = 34948)
        ISP: Zen Internet Ltd
    Latency:    14.63 ms   (0.27 ms jitter)
   Download:   916.69 Mbps (data used: 989.7 MB )
     Upload:   110.97 Mbps (data used: 90.0 MB )
Packet Loss:     0.0%
 Result URL: https://www.speedtest.net/result/c/57590364-51f6-4240-876f-01fc9c154103

C:\Users\ned\Downloads\sp>tracert www.google.co.uk

Tracing route to www.google.co.uk [2a00:1450:4009:81d::2003]
over a maximum of 30 hops:

  1    16 ms    <1 ms    <1 ms  OPNsense.localdomain ]
  2    14 ms    15 ms     9 ms  lo0-0.bng4.wh-man.zen.net.uk [2a02:8010::156]
  3    14 ms    15 ms    14 ms  lag-8.p2.wh-man.zen.net.uk [2a02:8010:0:900::1a]
  4    15 ms    14 ms    14 ms  ae-3.p1.thn-lon.zen.net.uk [2a02:8010:0:b00::3e]
  5    14 ms    14 ms    14 ms  2a00:1450:8136::1
  6     *       14 ms    14 ms  2001:4860:0:1::54ca
  7    14 ms    14 ms    14 ms  2001:4860:0:1::54c9
  8    14 ms    14 ms    13 ms  lhr25s31-in-x03.1e100.net [2a00:1450:4009:81d::2003]


IPV4 only via london
C:\Users\ned\Downloads\sp>speedtest -s 34948

   Speedtest by Ookla

     Server: Swish Fibre - London (id = 34948)
        ISP: Zen Internet Ltd
    Latency:     7.20 ms   (0.12 ms jitter)
   Download:   912.86 Mbps (data used: 831.9 MB )
     Upload:   110.52 Mbps (data used: 53.9 MB )
Packet Loss:     0.0%
 Result URL: https://www.speedtest.net/result/c/9f14e202-4160-45fc-a0b0-ad648e707f79

C:\Users\ned\Downloads\sp>tracert -4  www.google.co.uk

Tracing route to www.google.co.uk [142.250.187.195]
over a maximum of 30 hops:

  1    <1 ms    <1 ms    <1 ms  OPNsense.localdomain
  2     6 ms     6 ms     6 ms  vt1.cor1.lond2.ptn.zen.net.uk [51.148.72.23]
  3     7 ms     7 ms     6 ms  lag-8.p2.ixn-lon.zen.net.uk [51.148.73.206]
  4     6 ms     7 ms     6 ms  lag-2.p2.thn-lon.zen.net.uk [51.148.73.138]
  5     7 ms     7 ms     6 ms  lag-2.br1.thn-lon.zen.net.uk [51.148.73.167]
  6     7 ms     7 ms     7 ms  72.14.223.28
  7     9 ms     8 ms     8 ms  209.85.249.149
  8     6 ms     6 ms     6 ms  142.251.54.35
  9     6 ms     6 ms     6 ms  lhr25s33-in-f3.1e100.net [142.250.187.195]

[bogof]

Zen told me I am on a btw circuit and following a random reboot of my opnsense box a few days ago  I notice my latency shot up to almost double what it would normally be.. It now seems when ipv6 is enabled and I end up on lo0-0.bng4.wh-man.zen.net.uk I get around 14ms yet when I disable ipv6 and I seem to go via vt1.cor1.lond2.ptn.zen.net.uk using just ipv4 my latency goes back to 6/7ms..

Dont wish to hi-jack the thread but this just seems odd and has never been an issue before.

any thoughts?, why is the man gateway giving me a much greater latency?

I think this is fairly standard for Zen.  They have gateways in London and Manchester (3x lon and 2x man I believe that I've been able to count recently).  Additionally I think two of the London ones are at one physical location, and one is at a different one.  I didn't think there was an IPV6 trigger as to which you end up on - but I don't have IPV6 enabled.  Most of the time I tend to end up on a London gateway, but I will occasionally get put onto a Manchester one, with a resulting worsening of latency.  I think the Manchester gateways can be of benefit if you are closer to Manchester than London, but I'm not sure as that would depend on exactly how and where traffic flows at various points, and that's not really discussed openly anywhere.

In any case, you can probably expect to end up on a Manchester gateway occasionally with Zen.  Whether or not that's a problem for you I guess depends on your requirements.  It is a bit odd this variability in their network still exists (particularly given their gaming claims) but it is what it is. 

[skyeci]

Ok thanks. Seems that's a new thing to me as historically I have always connected to London which is closer than Manchester for me. Have messed about with it and 99% of the time I'm London with v4. Still can't get back on London with v6 as it keeps reverting to man which is annoying... 

[bogof]

Ok thanks. Seems that's a new thing to me as historically I have always connected to London which is closer than Manchester for me. Have messed about with it and 99% of the time I'm London with v4. Still can't get back on London with v6 as it keeps reverting to man which is annoying... 

I believe Zen have said elsewhere that whichever gateway answers first gets it; and that usually geography takes over so you will end up on the one closest; perhaps there is an issue with the London IPV6 connectivity at the moment, or something else. 

I had wondered myself  about enabling IPV6 and trying that, but I have limited IPV6 experience and didn't want to risk a change I didn't understand while trying to debug.

[skyeci]

My settings are pretty straight forward.
Wan ip settings dhcpv6
Request prefix
48
Use ipv4 connectivity
 On the lan side its set to track interface for ipv6 type.

Be interesting to see which gateway you get. When I asked Zen they just said they weren't aware of any problems with Manchester

[jelv]

London gateway and IPv6 working just fine for me:

Tracing route to google.co.uk [2a00:1450:4009:823::2003]
over a maximum of 30 hops:

  1     9 ms     3 ms     3 ms  fritz.box [2a02:8010:xxxx:0:7642:7fff:fe20:1e6c]
  2    18 ms    23 ms    59 ms  lo0-0.bng2.ixn-lon.zen.net.uk [2a02:8010::154]
  3    13 ms    12 ms    13 ms  2a02:8010::14a
  4    13 ms    13 ms    12 ms  ae-2.p1.thn-lon.zen.net.uk [2a02:8010:0:b00::3a]
  5    13 ms    12 ms    14 ms  2001:4860:0:135d::1
  6    12 ms    13 ms    13 ms  2001:4860:0:1::1a51
  7    13 ms    13 ms    12 ms  lhr48s30-in-x03.1e100.net [2a00:1450:4009:823::2003]

[skyeci]

So was mine till I rebooted 

[Chrysalis]

hmm

If Zen were marking speedtests as low priority, the obvious question would be why are they doing it?  My second issue with it is if anything an ISP would prioritise speedtests which is what they tended to do in the era when shaping was common in the UK.

I do agree of course though that just using a single speedtest as a means of diagnosing a connection isnt adequate, however the guys on the TBB thread including bogof in here have been doing their testing in sound ways and have repeatable behaviour when switching gateways, and switching between BTW and Zen backhaul of large performance differences, that kind of rules out typical internet backhaul or end user equipment causes.

[bogof]

hmm

If Zen were marking speedtests as low priority, the obvious question would be why are they doing it?  My second issue with it is if anything an ISP would prioritise speedtests which is what they tended to do in the era when shaping was common in the UK.

I do agree of course though that just using a single speedtest as a means of diagnosing a connection isnt adequate, however the guys on the TBB thread including bogof in here have been doing their testing in sound ways and have repeatable behaviour when switching gateways, and switching between BTW and Zen backhaul of large performance differences, that kind of rules out typical internet backhaul or end user equipment causes.

Indeed, and you'd have to hope that the Samknows box tests will represent their very best effort speed test results, as those results are what are used for the purposes of generating the advertised averages for the VCoP, so in a way it will be very interesting to see the Samknows box performance chart over time and across gateways.

Based on the current results I've seen with Samknows speedtest website, and going on what others have reported on their lines vs that test; assuming the Samknows box behaves similarly to what I'm seeing I'm not really expecting the Samknows box to hit line rate, even at off-peak times.  I think personally that would be a surprising result for a 900Mbps FTTP service, and indicative of something odd going on (even if the speed it hits is above the paltry 450Mbps Zen guarantee level). 

We shall see...

[skyeci]

I have applied for a Sam knows box. I'm interested to see what my 900 service is up too given I have a long time left on my contract...waiting to find out if I get one.

[craigski]

If Zen were marking speedtests as low priority, the obvious question would be why are they doing it? 

I'm only guessing here, but it makes sense to prioritise time sensitive traffic, such as video and VoIP. I would certainly hope that Netflix/BBC/Apple/Spotify/etc traffic had a higher priority on the Internet, than some potential DoS speedtest service.

I read the Samknows blog after J0hn posted, and their methodology is better. They are in control of when the speedtests run, and what servers, rather than ad-hoc tests on random servers that may or may not be busy.

The pings are also probably irrelevant. Responding to a ping should always be a lower priority than processing traffic.

My own experience, if I create network traffic from multiple devices (real world traffic, like downloading ISOs from say Microsoft) , I can nearly always see a higher utilisation on the WAN interface on my firewall vs trying to run a speedtest.

Has anyone tried my 2 minute experiment I posted on this thread a few pages back, ie try to prove your connection is faster that the speedtest result.

[Alex Atkin UK]

I'm only guessing here, but it makes sense to prioritise time sensitive traffic, such as video and VoIP. I would certainly hope that Netflix/BBC/Apple/Spotify/etc traffic had a higher priority on the Internet, than some potential DoS speedtest service.

While I agree in principle, in practice there are a lot of problems so the likes of AAISP and Zen DO NOT apply this as their networks are strictly complying to net neutrality.

The problem is it becomes a slippery slope once you start throttling as why should Netflix get priority when a smaller business offering similar services does not?  Its also another cost as you need much more powerful routers to handle this, especially at the speeds were dealing with today.  Plus another point of failure or misconfiguration.

Then there's the whole issue of court orders demanding you block sites, once you have the hardware to do so you MUST comply, whereas right now they can stand by it being too expensive due to needing huge network upgrades to achieve.

I originally left Plusnet because their traffic shapers were broken, they were supposed to throttle Usenet traffic after a specific amount of data per month, but was permanently throttling mine.  Then when they added site blocking I had a few occasions where instead of blocking the offending page, they blocked the whole domain.  Enough was enough at that point.

While they no longer apply any traffic shaping that I'm aware of, the latter remains.  Plus having that equipment means they are more likely to be logging your activity for the government.

I do not agree with Internet gatekeeping, as its also being used to "protect the children" which should NOT be an ISPs responsibility (as its unattainable and just encourages a false sense of security by parents), just as its not up to your telephone provider to decide which numbers you should and should not be allowed to call.  Its an Internet Service Provider, not a bits of the Internet we deem fit service provider.  The only thing they should be blocking are cyber attacks, where it would be service affecting if they didn't.

People get up in arms when councils make roads into bus/bike only, or block them off entirely, yet they're okay with someone deciding which parts of the Internet they are allowed on?  Its rather like your car deciding if you're driving to Sainsburys you must take a huge detour because your dealer has prioritised Tesco. (and as cars get smarter I fear this sort of crap WILL happen)

[craigski]

While I agree in principle, in practice there are a lot of problems so the likes of AAISP and Zen DO NOT apply this as their networks are strictly complying to net neutrality.

Net Neutrality is something totally different. I'm talking about prioritisation of specific time sensitive traffic, streaming, video calls, VoIP etc, eg:

https://www.aa.net.uk/voice-and-mobile/voip-information/

Works well with A&A broadband
Our own broadband services have constant quality monitoring and will give priority to VoIP traffic by default, ensuring good quality calls.

If the above is true, then A&A will drop 'speedtest' packets in favour of VoIP if there is congestion . I would assume other ISPs do similar.

[Alex Atkin UK]

Net Neutrality is something totally different. I'm talking about prioritisation of specific time sensitive traffic, streaming, video calls, VoIP etc

It literally isn't, by prioritising traffic you are breaking net neutrality, the whole premise is all traffic is treated equally.

It encourages the networks to maintain enough capacity, rather than just letting their network congest (common in the US) and prioritise the companies who are willing to pay for priority.