Milton Keynes routed via Manchester, high latency - Gigafast Broadband 900

[gerhardlazu]

My Gigafast Broadband 900 connection that is based in Milton Keynes is getting routed through Manchester instead of London. This means that all data packets go 120 miles north, and then 160 miles south before they can start getting routed to the correct destination.

Latency to bbc.co.uk (London area) is no less than 14ms, instead of 2-3ms:

$ ping -c 10 bbc.co.uk
PING bbc.co.uk (151.101.192.81): 56 data bytes
64 bytes from 151.101.192.81: icmp_seq=0 ttl=57 time=14.999 ms
...

--- bbc.co.uk ping statistics ---
10 packets transmitted, 10 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 14.359/14.643/15.033/0.213 ms

A service hosted in the Manchester area:

$ ping -c 10 ae5-100-xcr1.man.cw.net
PING ae5-100-xcr1.man.cw.net (195.89.96.113): 56 data bytes
64 bytes from 195.89.96.113: icmp_seq=0 ttl=60 time=9.904 ms
...

--- ae5-100-xcr1.man.cw.net ping statistics ---
10 packets transmitted, 10 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 9.194/9.976/10.692/0.382 ms

Did anyone else experience a similar behaviour? Did you manage to get it solved? How?

Thanks!

[MarkD]

Hi @gerhardlazu 

Best option could be to speak to directly to our Gigafast team who can advise you on this. It’s free to call us on 191 from your Vodafone mobile or 03333 040 191 from other UK landlines or mobiles (standard call charges apply). Lines are open Monday to Sunday 8am - 8pm.

Let us know how you get on 🙂

[gerhardlazu]

From previous experiences, it's the type of support query that goes nowhere. I'll give it a go though, thanks!

[clint_flick]

Hi

PING ms 461 DOWNLOAD Mbps 25.76 UPLOAD Mbps 0.86 Perth

PING ms 20 DOWNLOAD Mbps 33.99 UPLOAD Mbps 18.21 Lancaster

C:\>ping bbc.co.uk

Pinging bbc.co.uk [151.101.128.81] with 32 bytes of data:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),

Minimum = 20ms, Maximum = 20ms, Average = 20ms

C:\>tracert bbc.co.uk

Tracing route to bbc.co.uk [151.101.128.81]
over a maximum of 30 hops:

1 2 ms 2 ms 1 ms vodafone.broadband [192.168.1.1]
2 15 ms 16 ms 15 ms host-212-158-250-32.dslgb.com [212.158.250.32]
3 15 ms 15 ms 15 ms 63.130.140.126
4 19 ms 19 ms 19 ms 195.2.23.85
5 19 ms 19 ms 21 ms ae36-xcr1.lns.cw.net [195.2.25.170]
6 19 ms 19 ms 27 ms ae1-xcr1.ltw.cw.net [195.2.24.125]
7 19 ms 19 ms 19 ms 23.235.41.84
8 20 ms 21 ms 20 ms 151.101.128.81

Trace complete.

Maybe double check with another tracert?

But I am intrigued as to my Point Of Presence, where I enter and leave the network.

[Anonymous]

The dslgb network is mostly the interconnect between OpenReach and Vodafone. For the most part, it's rather opaque but it seems to be via a hub, rather than a dedicated link at the OpenReach local exchange level.  It's also worth noting that while it'll normally connect to the one data-center, if there are outages, it is diverted - with no apparent intermediate routing.

As for connecting to the bbc, the average ping from Teeside is 16ms, but it's worth being aware that there are fallbacks, parallel, and redundant servers in place, so we are almost certainly not all looking at the exact same server, possibly not even in the same location!

Regards the routing of Vodafone/CityFibre gigafast, it's going to be hard routed to a data centre.  At the moment that location for Milton Keynes would appear to the Manchester data centre.  That Manchester data centre has both connections to the UK backbone and to Vodafone/C&Ws parallel network, so everything that connects to Manchester does not then route through London (in fact very little international traffic does).  As for expecting a latency of 2-4ms latency on connections from Milton Keynes to London, those are unrealistic expectations, a congested local WiFi can add almost that much latency!

Pinging raspi2 [192.168.1.5] with 32 bytes of data:
Reply from 192.168.1.5: bytes=32 time=1ms TTL=64
Reply from 192.168.1.5: bytes=32 time=2ms TTL=64
Reply from 192.168.1.5: bytes=32 time=2ms TTL=64
Reply from 192.168.1.5: bytes=32 time=2ms TTL=64

[gerhardlazu]

Even though this is not the solution, as the forum software claims, it's the best answer with the most detail - thank you!

Here's an mttr of ~1300 pings to 216.58.205.35 (google.co.uk):

 Host                               Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. unifi.localdomain                0.0%  1294    0.1   0.1   0.1   0.4   0.0
 2. 192.168.101.1                    4.4%  1293    0.6   0.5   0.3   5.8   0.3
 3. 84.65.128.1                     21.1%  1293    9.5   9.4   8.1 200.2   7.3
 4. 63.130.127.221                  37.0%  1293   16.9  16.2  15.0  17.6   0.4
 5. 90.255.251.18                   36.0%  1293   16.2  16.7  15.2 100.6   3.8
 6. 74.125.242.97                   41.9%  1293   17.3  17.5  15.9  69.6   1.9
 7. 172.253.71.191                  36.0%  1293   16.8  16.4  15.3  17.3   0.3
 8. lhr48s23-in-f3.1e100.net        40.4%  1293   16.8  16.3  15.1  21.5   0.4

This shows:

• 0.6ms LAN latency (wired connection) - unifi.localdomain (Unifi router) & 192.168.101.1 (Vodafone modem)
• 9.4ms latency to 84.65.128.1, which is a Vodafone gateway in Crewe and adds the most latency (8.8ms) - this is the hop up to Manchester
• 16.2ms latency to 63.130.127.221, which is a hop inside the Vodafone network and adds the second most latency (6.2ms) - I am assuming this is the hop down to London
• 16.7ms latency to 90.255.251.18, also a hop inside the Vodafone network with minimal latency - the edge of the Vodafone network
• 17.5ms latency to 74.125.242.97, which is the entry point to Google's network in the London Heathrow area, most likely one of the Staines data centres

Notice how average latency decreases after packets enter the Google network, as does stddev. Vodafone Manchester gateway (84.65.128.1) has the most stddev, at 7.3ms, as does what I think is the Vodafone London gateway (90.255.251.18), at 3.8ms. They are by far the most congested hops, as hinted by the 200ms & 100ms slowest pings. The Google gateway is third most loaded, with 70ms slowest pings, but that's to be expected given the amount of traffic that it serves.

This is how packets travel currently:

Milton Keynes (Vodafone) -> Crewe (Vodafone) -> London (Vodafone) -> WHEREVER

This is how I would expect packets to travel:

Milton Keynes (Vodafone) -> London (Vodafone) -> WHEREVER

While 2-3ms is unrealistic, 5-6ms would be a reasonable expectation given the above measurements.

[clint_flick]

Hi

A decade ago, or a dozen years even, I may have been more helpful.....

https://www.cisco.com/c/en/us/td/docs/net_mgmt/prime/network/3-8/reference/guide/routpro.html

The routing protocols are kinda the metric for this. I thought OSPF myself.

SHORTEST PATH FIRST is an OPEN standard (OSPF) and in theory runs a stopwatch on the time to each/next hop and chooses the one with the lowest time, think Virgin Media and Usain Bolt, rather than Mo Farrer.

Since they have no need of human intervention once setup, the route your packets take is variable.

[gerhardlazu]

As a last update, a few weeks after I posted this, the issue fixed itself. Thank you. I'm currently averaging 5.9ms to google.co.uk:

                                     Packets       Pings
 Host                                Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. unifi.localdomain                 0.0%  1747    0.1   0.1   0.1  25.0   0.8
 2. 192.168.101.1                     6.8%  1747    0.7   0.6   0.3  11.3   0.4
 3. 84.65.0.1                        17.7%  1746    5.7   5.5   4.2  30.3   1.1
 4. 63.130.127.213                   37.6%  1746    6.1   5.8   4.4  13.9   0.5
 5. 90.255.251.18                    38.7%  1746    6.0   6.6   4.7 123.7   6.2
 6. 74.125.242.65                    38.6%  1746    5.0   6.2   4.8  13.6   0.7
 7. 172.253.50.223                   38.1%  1746    7.3   6.9   4.8  16.6   0.9
 8. dub08s01-in-f163.1e100.net       38.3%  1746    6.0   5.9   4.3  47.7   1.4

[purrbox]

Any more updates on this, I've been suffering the same issue, rerouted and high pings.

[purrbox]

.