Spark BNG seems to be overloaded

Forums › Spark New Zealand (including Skinny and BigPipe) › Spark BNG seems to be overloaded

p00t

5 posts

Wannabe Geek

#239856 8-Aug-2018 11:44

Does Spark regularly oversubscribe their BNGs?

I am doing a rather large download and was hoping to get more than 50 Mbit/s on my 100/20 service. Mind you 50 Mbit/s isn't all that bad from the US.

However I noticed that the Spark BNG has an average RTT of 8s, at 11am!! This isn't even the busy hour of the day.

Talkiet

4793 posts

Uber Geek

Trusted

#2069742 8-Aug-2018 11:49

No.

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.

Talkiet

4793 posts

Uber Geek

Trusted

#2069743 8-Aug-2018 11:51

For a little more detail, the BNGs and AS gateways don't consider ICMP particularly important. If the BNG was dropping traffic or delaying packets then you would see all further upstream hosts having issues as well.

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.

Sounddude

I fix stuff!
1928 posts

Uber Geek

Trusted

2degrees

Lifetime subscriber

#2069791 8-Aug-2018 12:22

What TalkieT said. Our BNG's are built to do the same

Would rather them route than reply to your ICMP :-)

sbiddle

30853 posts

Uber Geek

Retired Mod

Trusted

Biddle Corp

Lifetime subscriber

#2069793 8-Aug-2018 12:25

Plenty of other threads on here from people thinking their ISP is broken because their BNG either doesn't respond to pings or that their latency is high because ICMP is deprioritised and they are unsure how to read MTR results. These other threads discuss pretty extensively why things are configured this way.

Nothing is wrong with Spark or your connection.

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#2069795 8-Aug-2018 12:26

Sounddude:

What TalkieT said. Our BNG's are built to do the same

Would rather them route than reply to your ICMP :-)

oh hey, look at that... It's almost like it's an industry standard!

#include <std_disclaimer>

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

p00t

5 posts

Wannabe Geek

#2069857 8-Aug-2018 13:03

Fair enough about ICMP pings.

Let's try UDP then:

There is only one router than seems to be configured to de-prioritise ICMP or traceroute: 125-239-10-1-adsl.sparkbb.co.nz

The other routers upstream of the BNG seem ok about it.

michaelmurfy

meow
13260 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#2069874 8-Aug-2018 13:25

Working as intended.

Michael Murphy | https://murfy.nz
Referral Links: Quic Broadband (use R122101E7CV7Q for free setup)

Are you happy with what you get from Geekzone? Please consider supporting us by subscribing.
Opinions are my own and not the views of my employer.

Quic Broadband

Move to New Zealand's best fibre broadband service (affiliate link). Free setup code: R587125ERQ6VE. Note that to use Quic Broadband you must be comfortable with configuring your own router.

cbrpilot

955 posts

Ultimate Geek

Trusted

Spark NZ

#2069886 8-Aug-2018 13:49

Just because mtr is sending a UDP packet with a low TTL in the IP header instead of an ICMP packet, it does not mean that our BNG will generate a UDP reply.

It will still send an ICMP TTL Exceeded message back to the sender.

I.e. you're essentially testing the same thing.

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

p00t

5 posts

Wannabe Geek

#2070153 8-Aug-2018 16:56

I was looking more at the 7s delay time to process the TTL exceeded message. I wouldn't have thought that the Spark control plane would be buffering 7s worth of messages and so it would have been the network interface buffering causing the delay.

But of course then everything upstream would have had at least 7s of delay. Maybe I posted too soon...

Strangely the BNG is responding much faster to pings to it's interface than to TTL=0 frames:

BarTender

3606 posts

Uber Geek

ID Verified

Trusted

Lifetime subscriber

#2070167 8-Aug-2018 18:14

It's almost like you're complaining about a non issue.

yitz

2081 posts

Uber Geek

#2070186 8-Aug-2018 19:47

Just read it as an artefact.

ResponseMediaNZ

518 posts

Ultimate Geek

ID Verified

Trusted

#2070319 8-Aug-2018 22:09

Try listening to the guys in the tread that actually do this as a day job for the Telecos so might know a couple of things about how thierequipment works.

when you say you are not getting what you think you should be getting in download.. Are you on Wireless? No one else streaming netflix or gaming?

cbrpilot

955 posts

Ultimate Geek

Trusted

Spark NZ

#2070479 9-Aug-2018 10:10

@p00t can you advise what router you're using?

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

muppet

2570 posts

Uber Geek

Trusted

#2070493 9-Aug-2018 10:40

"But of course then everything upstream would have had at least 7s of delay."

This is exactly why what you've posted is a non-issue :)