Inexplicable difference in relative speed for national traffic over two UFB connections on the same exchange.

Forums › New Zealand Broadband › Inexplicable difference in relative speed for national traffic over two UFB connections on the same exchange.

dc2daylight

87 posts

Master Geek

#248881 14-Apr-2019 15:41

So I've switched over to Voyager fibre from Vodafone recently, and also switched my parents over. It was mostly a good choice, and the VOIP phonelines are working well now after some minor hurdles initially.

However I'm not all that impressed with speeds so far, and my place is about 6Mbps slower for linux updates with Mint to the Canterbury University mirror from Auckland, and is connected to the same street cabinet and exchange nearby. (I tested both connections at peak times, and neither are over Wifi.)

The only thing which could make a marginal difference on my end I'd like to rule out is that I'm connected through two switches between the Voyager supplied HG659 router, and the client pc in question. The slowest switch is 100Mbps and not intelligent/managed. I can't see how upstream packets could account for the 100Mbps switch causing the slowdown, so what else can I try?

As for speedtests, I have asked Voyager and been recommended to use Nperf instead of Speedtest or Speedof. I don't see how synthetic speedtests are relevant anyway and suspect my line is throttled for some reason, or else I'm not on the same cabinet afterall. I'm 100% it was the same cabinet when we were with Vodafone, and both connections were for xDSL instead.

*edited to change MBps to Mbps*

1 | 2

8955 posts

Uber Geek

#2216987 14-Apr-2019 15:50

So, what speed are you getting? 6 Mb/s slower than what? What connection speeds? What upstream packets are you referring to? Traceroute?

UFB isn't like xDSL in terms of being connected to a specific cabinet anyway, so forget about that.

RunningMan

8955 posts

Uber Geek

#2216992 14-Apr-2019 15:54

Going from your other posts, this is a Chorus connection - the only Chorus UFB connection that won't be throttled by a fast ethernet switch is 50/10.

dc2daylight

87 posts

Master Geek

#2216995 14-Apr-2019 16:15

RunningMan:
So, what speed are you getting? 6 Mb/s slower than what? What connection speeds? What upstream packets are you referring to? Traceroute?

UFB isn't like xDSL in terms of being connected to a specific cabinet anyway, so forget about that.

"So, what speed are you getting?" - Always less than 2.5Mbps to the Canterbury mirror no matter what time of day. The connection profile is 100/30 Fibre at both properties.

"6 Mb/s slower than what?" - Approximately 6Mb/s slower than my parents UFB connection, with the same exact operating system and version for 64bit on a different PC at their place. Theirs is usually 8Mb/s minimum.

"What upstream packets are you referring to?" - By upstream packets I meant acknowledge and syn packets.

"Traceroute?" When I do a traceroute it blackholes after every hop from my router outward, so there's no point posting it to here as no hostnames will come up nor time measurements. It does the same using tracepath instead of traceroute util also.

Thanks for clarifying about the UFB not being connected to a cabinet. I am not familiar with the network topology as I am very new to fibre in general.

*Edited to add connection profile*

gaddman

224 posts

Master Geek

Trusted

#2217017 14-Apr-2019 17:13

Can you test without your switches? Agree the speed ssd's shouldn't be a limiting factor, but maybe they're doing something funky adding latency. (I'm assuming the serialization delay is insignificant). Also, is there any difference in latency between you and your parents?

RunningMan

8955 posts

Uber Geek

#2217019 14-Apr-2019 17:22

What's the mirror URL? 8 Mb/s suggests it's a limitation of the server, as even that is nowhere near the limits of your parent's connection.

TheoM

228 posts

Master Geek

ID Verified

Trusted

#2217074 14-Apr-2019 19:57

RunningMan:

What's the mirror URL? 8 Mb/s suggests it's a limitation of the server, as even that is nowhere near the limits of your parent's connection.

The mirror URL would be http://ucmirror.canterbury.ac.nz/linux/

I'm sure it's his connection and not UC, last I checked they impose a hard 15MB/s (120Mbps) cap on any one connection, which I may or may not have had a hand in causing *ahem*

In any case, UC is not the limiting factor, it's more likely to be @dc2daylight's connection.

Hi! I'm TheoM, but you know that already. I run Linux mirrors in NZ together with 2degrees. Like a mirror added? PM me!

https://theom.co.nz | https://theom.nz | https://mirrorlist.mirrors.theom.nz | Providing Free Mirrors Since Ages Ago™

RunningMan

8955 posts

Uber Geek

#2217084 14-Apr-2019 20:16

That's the URL I assumed - I'm wondering if that's actually 8 MB/s.

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.

dc2daylight

87 posts

Master Geek

#2217104 14-Apr-2019 21:23

gaddman: Can you test without your switches? Agree the speed ssd's shouldn't be a limiting factor, but maybe they're doing something funky adding latency. (I'm assuming the serialization delay is insignificant). Also, is there any difference in latency between you and your parents?

I've already tested once without the switches in the way and there was no difference, but will try it again tomorrow and post the results if different, which I highly doubt they will be other than a few Kbps.

As for latency at my parents, I'll also try a traceroute up there tomorrow and see if it goes through hosts with no ICMP or UDP replies/blackholed.

The other strange aspect is that when I launch a google maps session my default location shifts around randomly to ones outside auckland, including when allowing location tracking in firefox. My connection has no funky routing or ISP side VPN that I'm aware of, and should act like anyone elses. Standard residential customer here, that's it...

Just tried a traceroute to geekzone and I get the same lack of results, so something weird is up with the routing alright.

*edit to add info about further traceroutes*

It happens with all traceroutes now, so I'll presume someone at chorus or voyager has disabled ICMP tracroutes on my connection. What a joke...not.

*edit to add facepalms-akimbo*

Ok sorry my bad, I had the firewall blocking ICMP outward on the Huawei. I can now do traceroutes.

traceroute to ucmirror.canterbury.ac.nz (132.181.7.179), 30 hops max, 60 byte packets

1 192.168.1.1 (192.168.1.1) 0.588 ms 0.684 ms 0.790 ms

2 114.23.3.255 (114.23.3.255) 3.703 ms 3.754 ms 3.747 ms

3 114.23.3.230 (114.23.3.230) 34.789 ms 6.127 ms 6.332 ms

4 as23655.akl-ix.nz (43.243.21.30) 6.341 ms 6.302 ms 6.314 ms

5 161.30.69.111.static.snap.net.nz (111.69.30.161) 25.236 ms 25.131 ms 25.212 ms

6 163.30.69.111.static.snap.net.nz (111.69.30.163) 39.931 ms 36.629 ms 36.023 ms

7 202.36.179.100 (202.36.179.100) 22.450 ms 22.156 ms 21.944 ms

8 132.181.3.236 (132.181.3.236) 22.321 ms 22.194 ms 22.373 ms

9 132.181.7.179 (132.181.7.179) 22.080 ms !X 22.164 ms !X 21.907 ms !X

dc2daylight

87 posts

Master Geek

#2220111 17-Apr-2019 18:55

1) I tried without the switches in the way, and there's no difference as expected. Mint updates to a 3rd and different machine this afternoon (plugged directly into the HG659 over Cat5) were at the same hard limit of approx 2.7Mbps this time, using the Canterbury server.

2) On testing with both netperf.com and testmy.net (using the sydney server single-threaded) my parents connection is way faster than mine at 69.1Mbps at 05:20pm yesterday. This is about the speed I'd expect my own Voyager fibre to be ideally and realistically. Mine was a piddly 20.9Mbps at 04:35pm yesterday with the same settings on testmy.net.

3) The traceroute to the server is identical in the hops from both physical sites and the timing variation is minimal, but I do have a screenshot if it helps.

So is this likely to be down to the exchange that my Chorus wholesale UFB is connected to being overloaded, or will the problem be with Voyager?

RunningMan

8955 posts

Uber Geek

#2220129 17-Apr-2019 19:21

I would suggest a factory reset of your router, and try again.

dc2daylight

87 posts

Master Geek

#2220255 17-Apr-2019 22:28

RunningMan:
I would suggest a factory reset of your router, and try again.

Are you *expletive* joking? I hope so... In fact I'd say so, so fair enough for your merriment.

michaelmurfy

meow
13245 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#2220268 17-Apr-2019 23:55

dc2daylight:
RunningMan:

I would suggest a factory reset of your router, and try again.

Are you *expletive* joking? I hope so... In fact I'd say so, so fair enough for your merriment.

He is serious and honestly I've seen this resolve many things here. There is a reason why ISP's ask you to do things like this (rebooting, factory resetting etc). And to add more to this - don't just do a factory reset and restore your configuration. Actually do a factory reset, ensure you're running the latest firmware (on the HG659) and also set it up from scratch to ensure you have not got any potentially dodgy stuff in the configuration.

Testing on my parents Voyager VDSL connection I can almost max full line speed to the Canterbury Linux mirror. As a Linux user myself, I've never at all had any issues with this.

I'm suspecting the problem may be with your router. The HG659 is great at Gigabit however it is only great with Gigabit with smaller homes. It sounds like you've got a fair bit connected to it. There are way too many variables you're putting in the mix to compare your parents and your own connection. I can (at a guess) imagine you've got way more devices connected to your network and your routers routing table will have a fair chunk more in it also.

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.

RunningMan

8955 posts

Uber Geek

#2220696 18-Apr-2019 16:09

dc2daylight:
RunningMan:

I would suggest a factory reset of your router, and try again.

Are you *expletive* joking? I hope so... In fact I'd say so, so fair enough for your merriment.

No, 100% serious.

The settings have clearly been mucked with if ICMP was blocked by the firewall and you didn't realise, so it's reasonable to assume that other settings have also been changed which are affecting your connection.

dc2daylight

87 posts

Master Geek

#2220758 18-Apr-2019 17:24

It's also reasonable to assume I may actually have a clue what I'm doing. I set the firewall to high as I don't like packets bouncing around my local network for services I don't need or use, and found the wording in the Huawei inbuilt help/explanations to be ambiguous about whether they really meant nothing but HTTP,FTP or DNS. A good example of this ambiguity is that NTP is allowed and works just fine with the "firewall" on "high".

dc2daylight

87 posts

Master Geek

#2220767 18-Apr-2019 17:52

michaelmurfy:
dc2daylight:
RunningMan:

I would suggest a factory reset of your router, and try again.

Are you *expletive* joking? I hope so... In fact I'd say so, so fair enough for your merriment.

He is serious and honestly I've seen this resolve many things here. There is a reason why ISP's ask you to do things like this (rebooting, factory resetting etc). And to add more to this - don't just do a factory reset and restore your configuration. Actually do a factory reset, ensure you're running the latest firmware (on the HG659) and also set it up from scratch to ensure you have not got any potentially dodgy stuff in the configuration.

Testing on my parents Voyager VDSL connection I can almost max full line speed to the Canterbury Linux mirror. As a Linux user myself, I've never at all had any issues with this.

I'm suspecting the problem may be with your router. The HG659 is great at Gigabit however it is only great with Gigabit with smaller homes. It sounds like you've got a fair bit connected to it. There are way too many variables you're putting in the mix to compare your parents and your own connection. I can (at a guess) imagine you've got way more devices connected to your network and your routers routing table will have a fair chunk more in it also.

Naturally I had already performed a factory reset, and in fact had to because of a mixup with provisioning about which device was provided for which property and me being given mixed messages about whether the WAN side MAC address would affect VOIP having the correct allocated ported phone number. I was told by one CSR it would not matter, and another agreed with me during another support call that yes it did seem to matter. Mistakes happen with support teams, just like by customers eg me. No one is perfect.

I've set up dozens of routers for many people over the past 2 decades, and one particular domestic model TP-Link I set up (a great purchase thanks to Cyrils excellent posts here over the years which I have learnt a lot from) stayed operational for around 30 or more Wifi Clients with all manner of dodgy devices coming and going, and all manner of dodgy traffic over a struggling slingshot ADSL2+ connection over about 5 years with no reset being needed. Putting Client Isolation on probably helped a lot! As did turning off everything but the barebones network services. One time I had to sniff traffic with wireshark to see who's PC was pumping out DNS redirections, and it turned out to be an HP Laptop with a buggy "Network Diagnostics" bloatware offering. Uninstalling that software for the resident of the house, with his permission, returned the router back to an operational state again. Traffic had slowed to a crawl before the fix.

But anyway returning to the purpose of this thread - The resets I've already done made zero difference. When I set up the device in my place for the first time I spent far too long poring over every single setting in the router and would have done so by SSH or Telnet with true root admin rights if I had had them, because it's my place and my rented equipment, and because voyager provide clearly written settings in their provisioning emails for the accounts. The HG659 seems to be running older firmware from what I recall, and seeing as it's on a leased agreement and remains Voyagers property I won't be swapping out the firmware anytime soon. I left TR-069 on for the sake of cooperation too.

As for my setup here - There's only two active PCs in use currently, and I restricted DHCP to less than 20 possible clients in the IP address range, as I have always done that and it lets routers run cooler and happier in my experience so I doubt the routing table is the issue. But thanks for the educated guess. All wifi is usually off, and when it is on there would only ever be 1-2 devices in use at any time, and if something slowed down my network to a cludge I'd find the device and block the MAC addresses if unauthorized somehow.

Both PCs I tested with have gigabit LAN ports, and are running linux solely so drivers aren't the issue unless it's the kernel version at fault.

I think though you may all have a point about the fault being with the router, so I'll try with my old Vodafone Huawei after wiping out Vodafone specific settings, and I'll also try my spare Draytek I have lying around and let this thread know what happens. Thanks for trying to help, and if I can't get a better result I will then lodge a support ticket with Voyager.

1 | 2