BigPipe high latency fibre connection

Forums › Spark New Zealand (including Skinny and BigPipe) › BigPipe high latency fibre connection

1 | 2 | 3 | 4

BarTender

3629 posts

Uber Geek
+1 received by user: 2572

ID Verified

Trusted

Lifetime subscriber

#2522102 13-Jul-2020 14:21

cbrpilot:

Quick update from me here.

I believe we have found the issue for both Napier and Kapiti for Bigpipe customers. Thanks for bringing this to my attention.

It would appear that (unknown to us) Chorus seem to take your traffic on a grand tour before it is handed over to our network - and is working as designed from their perspective. Obviously not ideal and we are exploring options to see if we can avoid this. Can't make any guarantees at this point but we are working on it!

Dave.

This has made me smile (sort of).

Would be interesting to know if all other RSPs are impacted or if this is special to Spark.

linw

2893 posts

Uber Geek
+1 received by user: 1205

#2522110 13-Jul-2020 14:35

These are my pings from Porirua on Orcon fibre 100/20

ping 219.88.188.70

Pinging 219.88.188.70 with 32 bytes of data:
Reply from 219.88.188.70: bytes=32 time=22ms TTL=120
Reply from 219.88.188.70: bytes=32 time=22ms TTL=120
Reply from 219.88.188.70: bytes=32 time=22ms TTL=120
Reply from 219.88.188.70: bytes=32 time=23ms TTL=120

Ping statistics for 219.88.188.70:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 22ms, Maximum = 23ms, Average = 22ms

ping 219.88.188.134

Pinging 219.88.188.134 with 32 bytes of data:
Reply from 219.88.188.134: bytes=32 time=28ms TTL=120
Reply from 219.88.188.134: bytes=32 time=27ms TTL=120
Reply from 219.88.188.134: bytes=32 time=27ms TTL=120
Reply from 219.88.188.134: bytes=32 time=28ms TTL=120

Ping statistics for 219.88.188.134:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 27ms, Maximum = 28ms, Average = 27ms

ping 219.88.188.7

Pinging 219.88.188.7 with 32 bytes of data:
Reply from 219.88.188.7: bytes=32 time=13ms TTL=120
Reply from 219.88.188.7: bytes=32 time=14ms TTL=120
Reply from 219.88.188.7: bytes=32 time=11ms TTL=120
Reply from 219.88.188.7: bytes=32 time=11ms TTL=120

Ping statistics for 219.88.188.7:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 11ms, Maximum = 14ms, Average = 12ms

Talkiet

4819 posts

Uber Geek
+1 received by user: 3934

Trusted

#2522111 13-Jul-2020 14:39

The server IPs I gave to ping are only really a valid test measure for Spark/Bigpipe/Skinny customers. Other ISPs will be able to see them but are never served from them - so honestly I have never even thought about how optimised paths to them from other ISPs might be.

Cheers - N

Please note all comments are from my own brain and don't necessarily represent the position or opinions of my employer, previous employers, colleagues, friends or pets.

linw

2893 posts

Uber Geek
+1 received by user: 1205

#2522158 13-Jul-2020 14:51

Yea, wondered about that. Never mind.

BarTender

3629 posts

Uber Geek
+1 received by user: 2572

ID Verified

Trusted

Lifetime subscriber

#2522219 13-Jul-2020 15:32

EnragedStoat:

Tracing route to cache.google.com [219.88.188.70]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.81.1
2 18 ms 18 ms 19 ms 125-239-203-1.jetstream.xtra.co.nz [125.239.203.1]
3 21 ms 21 ms 21 ms 222.152.41.187
4 21 ms 21 ms 21 ms cache.google.com [219.88.188.70]

Trace complete.

The only way anyone else would see this is by doing a traceroute on a wired connection (not over wiif) and seeing the latency on the first hop after your router.

The issue is that Spark do de-prioritize ICMP traffic, but this does show up the problem.

This does make me chuckle a lot, and I wonder if other ISPs have the same / similar problem depending on how distributed their BNGs are across the country or if they are all aggregated back to Auckland.

JonoNZ

274 posts

Ultimate Geek
+1 received by user: 43

#2522371 13-Jul-2020 17:52

cbrpilot:

Quick update from me here.

I believe we have found the issue for both Napier and Kapiti for Bigpipe customers. Thanks for bringing this to my attention.

It would appear that (unknown to us) Chorus seem to take your traffic on a grand tour before it is handed over to our network - and is working as designed from their perspective. Obviously not ideal and we are exploring options to see if we can avoid this. Can't make any guarantees at this point but we are working on it!

Dave.

Good to hear, but I changed from BigPipe partly because of this, wouldn’t mind a contribution from Spark/BigPipe on the early termination charge as I was the reporter of this unexpected behaviour (despite being shot down by many!). Any tips on who I can contact there, BigPipe CS is not very helpful tbh, get what you pay for I guess.

I can confirm the Spark performs much better for those that care about latency in Napier. Night and day.

Samsung

Shop now on Samsung phones, tablets, TVs and more (affiliate link).

EnragedStoat

14 posts

Geek
+1 received by user: 3

#2522564 14-Jul-2020 08:46

cbrpilot:

I believe we have found the issue for both Napier and Kapiti for Bigpipe customers.

I would say so :) 1.0 x 10^3 thanks!

Pinging 219.88.188.70 with 32 bytes of data:
Packets: Sent = 12, Received = 12, Lost = 0 (0% loss),
Minimum = 2ms, Maximum = 2ms, Average = 2ms
(was previously Minimum = 21ms, Maximum = 21ms, Average = 21ms)

Pinging 219.88.188.134 with 32 bytes of data:
Packets: Sent = 11, Received = 11, Lost = 0 (0% loss),
Minimum = 6ms, Maximum = 7ms, Average = 6ms
(was previously Minimum = 25ms, Maximum = 26ms, Average = 25ms)

Pinging 219.88.188.7 with 32 bytes of data:
Packets: Sent = 11, Received = 11, Lost = 0 (0% loss),
Minimum = 11ms, Maximum = 11ms, Average = 11ms
(was previously Minimum = 30ms, Maximum = 32ms, Average = 30ms)

Also seeing improved throughput, previously I was lucky to hit 600Mbp/s, now consistently >900.

Assuming you can roll this fix out to other affected locations will it also work if I change back to a static IP?

hio77

'That VDSL Cat'
13036 posts

Uber Geek
+1 received by user: 3896

ID Verified

Trusted

Lizard Networks

Subscriber

#2522650 14-Jul-2020 12:19

JonoNZ:

Good to hear, but I changed from BigPipe partly because of this, wouldn’t mind a contribution from Spark/BigPipe on the early termination charge as I was the reporter of this unexpected behaviour (despite being shot down by many!). Any tips on who I can contact there, BigPipe CS is not very helpful tbh, get what you pay for I guess.

I can confirm the Spark performs much better for those that care about latency in Napier. Night and day.

Hey Jono,

we certainly appreciate it.

The team will be in contact :)

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

cbrpilot

964 posts

Ultimate Geek
+1 received by user: 555

Trusted

Spark NZ

#2522678 14-Jul-2020 13:43

EnragedStoat:

cbrpilot:

I believe we have found the issue for both Napier and Kapiti for Bigpipe customers.

I would say so :) 1.0 x 10^3 thanks!

Pinging 219.88.188.70 with 32 bytes of data:
Packets: Sent = 12, Received = 12, Lost = 0 (0% loss),
Minimum = 2ms, Maximum = 2ms, Average = 2ms
(was previously Minimum = 21ms, Maximum = 21ms, Average = 21ms)

Pinging 219.88.188.134 with 32 bytes of data:
Packets: Sent = 11, Received = 11, Lost = 0 (0% loss),
Minimum = 6ms, Maximum = 7ms, Average = 6ms
(was previously Minimum = 25ms, Maximum = 26ms, Average = 25ms)

Pinging 219.88.188.7 with 32 bytes of data:
Packets: Sent = 11, Received = 11, Lost = 0 (0% loss),
Minimum = 11ms, Maximum = 11ms, Average = 11ms
(was previously Minimum = 30ms, Maximum = 32ms, Average = 30ms)

Also seeing improved throughput, previously I was lucky to hit 600Mbp/s, now consistently >900.

Assuming you can roll this fix out to other affected locations will it also work if I change back to a static IP?

Hi EnragedStoat, we rolled this out solely to you as a test to ensure that it had the desired impact - which it has :)

We have not rolled this out more widely as yet (few hoops to jump through etc).

Assuming that goes smoothly we will endeavour to roll this out more widely in the coming weeks.

My views are my own, and may not necessarily represent those of my employer.

JonoNZ

274 posts

Ultimate Geek
+1 received by user: 43

#2522713 14-Jul-2020 14:36

hio77:

JonoNZ:

Good to hear, but I changed from BigPipe partly because of this, wouldn’t mind a contribution from Spark/BigPipe on the early termination charge as I was the reporter of this unexpected behaviour (despite being shot down by many!). Any tips on who I can contact there, BigPipe CS is not very helpful tbh, get what you pay for I guess.

I can confirm the Spark performs much better for those that care about latency in Napier. Night and day.

Hey Jono,

we certainly appreciate it.

The team will be in contact :)

Thanks, they’ve been in touch — much appreciated!

1 | 2 | 3 | 4