Manchester now seemingly routing via London

Not sure if anyone else can replicate this, but it seems that Hyperoptic has made some fairly significant core network changes.

For those in the North, it used to be the case that if Hyperoptic had peering at Linx Manchester, they'd use that for outbound traffic, failing that GTT or Zayo from Manchester. They'd still announce in Linx London / Lonap, for inbound traffic.

Lately it seems that unless going via Linx Manchester, they backhaul everything down to London and break out from there to either Linx, Lonap or transit (GTT and Zayo again).


I wonder what prompted this change?

Hello blueacid! Thank you for posting this question.
Changes like these are usually made to improve the service we provide, the network stability and to optimize service distribution.
If you believe that your service is in any way affected, let us know and we will engage our Network Team to check this information and monitor your connection. Thank you for reaching out!

Hey there,

Thank you for your query. We've consulted our Core Network team about this matter. Our team confirmed changes have been made in order to improve our customers' experience.

We hope this helps.

I am from Manchester and just about to apply for HO service. Is this routing change affecting much on ping time? Since I often do local connections within the North, it will be a shame if without links straight into Linx Manchester but routing down London and back.

By the way, what does the ping time look like for HO connection between Manchester and London?

To be honest, really not.
Without a network map or anything, I can't be 100% sure if this is correct, but it seems that from here there are several routes out.

1) IX Manchester. All the possible peers are here: https://www.peeringdb.com/ix/583 - I'm fairly sure that Hyperoptic peer with Amazon, Apple, Cloudflare, Akamai, Fastly, BBC etc at least. So a lot of websites and services are phenomenally low in ping:

~ ping -c 3 1.1.1.1
PING 1.1.1.1 (1.1.1.1) 56(84) bytes of data.
64 bytes from 1.1.1.1: icmp_seq=1 ttl=57 time=1.22 ms
64 bytes from 1.1.1.1: icmp_seq=2 ttl=57 time=1.30 ms
64 bytes from 1.1.1.1: icmp_seq=3 ttl=57 time=1.24 ms

--- 1.1.1.1 ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2003ms
rtt min/avg/max/mdev = 1.224/1.255/1.299/0.031 ms

ping -c 3 s3.eu-west-1.amazonaws.com
PING s3.eu-west-1.amazonaws.com (52.218.105.178) 56(84) bytes of data.
64 bytes from s3-eu-west-1.amazonaws.com (52.218.105.178): icmp_seq=1 ttl=42 time=16.0 ms
64 bytes from s3-eu-west-1.amazonaws.com (52.218.105.178): icmp_seq=2 ttl=42 time=15.8 ms
64 bytes from s3-eu-west-1.amazonaws.com (52.218.105.178): icmp_seq=3 ttl=42 time=16.3 ms

--- s3.eu-west-1.amazonaws.com ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2003ms
rtt min/avg/max/mdev = 15.848/16.022/16.250/0.168 ms



2) Down to London and then out from there to Linx, Lonap, Zayo and GTT
The pings to London really aren't slow - they seem to add around 6msec.
~ traceroute aaisp.net
traceroute to aaisp.net (81.187.30.81), 30 hops max, 60 byte packets
1 192.168.34.254 (192.168.34.254) 0.481 ms 0.795 ms 1.138 ms
2 88.98.247.xxx.bcube.co.uk (88.98.247.xxx) 8.884 ms 13.059 ms 13.627 ms
3 172.16.23.106 (172.16.23.106) 6.467 ms 6.753 ms 6.967 ms
4 172.16.24.92 (172.16.24.92) 14.564 ms 15.063 ms 172.17.7.84 (172.17.7.84) 7.155 ms
5 172.16.30.164 (172.16.30.164) 9.036 ms 9.627 ms 12.804 ms
6 172.16.28.10 (172.16.28.10) 15.414 ms 9.575 ms 11.705 ms
7 * * *
8 a.aimless.thn.aa.net.uk (90.155.53.41) 9.590 ms 10.106 ms 10.348 ms
9 46.249.195.227 (46.249.195.227) 12.625 ms 13.066 ms 10.460 ms
10 www-server.co.uk (81.187.30.81) 10.817 ms 10.157 ms 10.130 ms



In real terms, web browsing is very quick, downloads hoof along at a phenomenal pace, Google Meet, Hangouts, Microsoft Teams, Slack, and Zoom video conferences work well (These are all the ones I've used lately and am happy to confirm about - put it this way, nothing's NOT worked well!), and when I'm gaming, I usually have a lower ping than my mates on Virgin Media, Sky, and BT. Shame I'm not as good at COD as some hyperactive teenager that's full of sugar, but you can't win 'em all.
Going further afield, a friend hosts a private game server on OVH in France. When we're gaming, my ping to that is around 11-12msec.

I've not really bothered to run speedtests, because the internet is now at the same point that the electricity and water supplies are - it's just 'there' and I use it without worrying about whether it's going to be alright or if it'll cope.

If you do want to get the best out of your connection, I'd strongly suggest that you run ethernet cables to everything that's staying in one place. Yes, fitting uprated wifi points does definitely help, but even then they're no match for a dedicated network cable. This applies even on ADSL, mind you, so it's not hyperoptic-specific!

I hope that's helped, if you've got any hosts or addresses you want me to traceroute or ping, let me know!

In reply to a post by blueacid:

To be honest, really not.
Without a network map or anything, I can't be 100% sure if this is correct, but it seems that from here there are several routes out.

1) IX Manchester. All the possible peers are here: https://www.peeringdb.com/ix/583 - I'm fairly sure that Hyperoptic peer with Amazon, Apple, Cloudflare, Akamai, Fastly, BBC etc at least. So a lot of websites and services are phenomenally low in ping:

~ ping -c 3 1.1.1.1
PING 1.1.1.1 (1.1.1.1) 56(84) bytes of data.
64 bytes from 1.1.1.1: icmp_seq=1 ttl=57 time=1.22 ms
64 bytes from 1.1.1.1: icmp_seq=2 ttl=57 time=1.30 ms
64 bytes from 1.1.1.1: icmp_seq=3 ttl=57 time=1.24 ms

--- 1.1.1.1 ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2003ms
rtt min/avg/max/mdev = 1.224/1.255/1.299/0.031 ms

ping -c 3 s3.eu-west-1.amazonaws.com
PING s3.eu-west-1.amazonaws.com (52.218.105.178) 56(84) bytes of data.
64 bytes from s3-eu-west-1.amazonaws.com (52.218.105.178): icmp_seq=1 ttl=42 time=16.0 ms
64 bytes from s3-eu-west-1.amazonaws.com (52.218.105.178): icmp_seq=2 ttl=42 time=15.8 ms
64 bytes from s3-eu-west-1.amazonaws.com (52.218.105.178): icmp_seq=3 ttl=42 time=16.3 ms

--- s3.eu-west-1.amazonaws.com ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2003ms
rtt min/avg/max/mdev = 15.848/16.022/16.250/0.168 ms



2) Down to London and then out from there to Linx, Lonap, Zayo and GTT
The pings to London really aren't slow - they seem to add around 6msec.
~ traceroute aaisp.net
traceroute to aaisp.net (81.187.30.81), 30 hops max, 60 byte packets
1 192.168.34.254 (192.168.34.254) 0.481 ms 0.795 ms 1.138 ms
2 88.98.247.xxx.bcube.co.uk (88.98.247.xxx) 8.884 ms 13.059 ms 13.627 ms
3 172.16.23.106 (172.16.23.106) 6.467 ms 6.753 ms 6.967 ms
4 172.16.24.92 (172.16.24.92) 14.564 ms 15.063 ms 172.17.7.84 (172.17.7.84) 7.155 ms
5 172.16.30.164 (172.16.30.164) 9.036 ms 9.627 ms 12.804 ms
6 172.16.28.10 (172.16.28.10) 15.414 ms 9.575 ms 11.705 ms
7 * * *
8 a.aimless.thn.aa.net.uk (90.155.53.41) 9.590 ms 10.106 ms 10.348 ms
9 46.249.195.227 (46.249.195.227) 12.625 ms 13.066 ms 10.460 ms
10 www-server.co.uk (81.187.30.81) 10.817 ms 10.157 ms 10.130 ms



In real terms, web browsing is very quick, downloads hoof along at a phenomenal pace, Google Meet, Hangouts, Microsoft Teams, Slack, and Zoom video conferences work well (These are all the ones I've used lately and am happy to confirm about - put it this way, nothing's NOT worked well!), and when I'm gaming, I usually have a lower ping than my mates on Virgin Media, Sky, and BT. Shame I'm not as good at COD as some hyperactive teenager that's full of sugar, but you can't win 'em all.
Going further afield, a friend hosts a private game server on OVH in France. When we're gaming, my ping to that is around 11-12msec.

I've not really bothered to run speedtests, because the internet is now at the same point that the electricity and water supplies are - it's just 'there' and I use it without worrying about whether it's going to be alright or if it'll cope.

If you do want to get the best out of your connection, I'd strongly suggest that you run ethernet cables to everything that's staying in one place. Yes, fitting uprated wifi points does definitely help, but even then they're no match for a dedicated network cable. This applies even on ADSL, mind you, so it's not hyperoptic-specific!

I hope that's helped, if you've got any hosts or addresses you want me to traceroute or ping, let me know!

It looks like if the connections are within the North it's will still routing via Linx Manchester, rest going down London then elsewhere. This is what I want!

Do you mind to do some ping or trace route to Europe (Amsterdam?) and US (New York)? Just worrying if there is less bandwidth to overseas because it's cheaper than other ISP...

Grabbed a few locations from the CentOS mirrors list:https://www.centos.org/download/mirrors/
It's probably worth pointing out that I'm on the 150/150 package, so flat out for me is around 20MB/sec downstream.

Los Angeles: http://mirror.centos.lax1.serverforge.org/
Ping: 136msec
Speed: 18.8MB/sec

New York: http://mirrors.rit.edu/centos/
Ping: 73msec
Speed: 19.2MB/sec

Amsterdam: http://mirror.ams1.nl.leaseweb.net/centos/
Ping: 12.2ms
Speed: 19.2MB/sec


Yep, line rate on all transfers, ping times were consistent, international connectivity is definitely well sorted right now!