Congestion returning with a vengeance?

Forums › One New Zealand (including Vodafone, WxC, Farmside) › Congestion returning with a vengeance?

1 | ... | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | ... | 34

1209 posts

Uber Geek

ID Verified

Trusted

#1806373 25-Jun-2017 13:17

hio77: Remember with steam, since steampipe content server numbers are way down.

In nz all we have now are akamai based chaches (which often do an amazing job)
Thing to remember about this is the load balancing algorithm etc has a habbit of throwing you back to vavles main servers in the worst case.

With sales caches always struggle to stay loaded with apporate content for 99% of people.

Thanks for a great explanation. However when he compared to friends above the performance of fibreX seems sub par. But Mike Hayles says no network issues. Is this explainable because Steam seems to be on of the things people are pointing to to say that their 'real world performance' is crap.

Kirk

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#1806374 25-Jun-2017 13:22

kharris:
Thanks for a great explanation. However when he compared to friends above the performance of fibreX seems sub par. But Mike Hayles says no network issues. Is this explainable because Steam seems to be on of the things people are pointing to to say that their 'real world performance' is crap.

To take that explanation from another angle; comparing a less popular isp (myrp in the posters case) their akamai servers are probably being hit less.

Compare that to murfys results which last i check bigpipe ran off tg hosted caches predominantly. That's likely being hit by atleast 2 large isps along with a fair few smaller ones.
Scaling cache servers is a horrible task; i feel sorry for the network admins managing them.

Likely constrain right now I'd expect to be random io. Nothing bandwidth related. But when your customers are all pulling random games off them the contention will hit.

Sit back and have a cold one. The bandwidth will return soon.

#include <std_disclaimer>

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

kharris

1209 posts

Uber Geek

ID Verified

Trusted

#1806384 25-Jun-2017 13:54

Thanks Troy. So not likely to be a fibteX issue then, but unfortunately a big affect on the customer experience. I don't use Steam so not an issue for me. I just find this to be a really interesting discussion. It seems easy to lay blame initially. Sounds like he may be better off on MR based on his usage.

Kirk

Kodiack

709 posts

Ultimate Geek

Trusted

#1806400 25-Jun-2017 14:47

Needing to update my GPU drivers while troubleshooting an issue. Even this is painfully slow:

richms

28168 posts

Uber Geek

Trusted

Lifetime subscriber

#1806410 25-Jun-2017 14:54

On voyager I was maxing out my pathetic 20something megabit vdsl line on steam downloads. Its strange that even with the sale on, the only people that seem to be having problems with steam are those on a certain red ISP on the mislabeled to the point of an investigation network.

Perhaps steam is something that truenet need to add to their tests?

Richard rich.ms

gaddman

224 posts

Master Geek

Trusted

#1806441 25-Jun-2017 15:59

Kodiack:

Needing to update my GPU drivers while troubleshooting an issue. Even this is painfully slow:

I can't remember if you logged a fault already, but assuming you did is it still under investigation or have you been advised that it's fixed? I just tested across a few FibreX connections and got 20-50MBps, eg this FibreX Max in Wellington:

--2017-06-25 15:52:45-- http://us.download.nvidia.com/Windows/382.53/382.53-desktop-win10-64bit-international-whql.exe
Resolving us.download.nvidia.com (us.download.nvidia.com)... 192.229.232.112, 2606:2800:247:2063:46e:21d:825:102e
Connecting to us.download.nvidia.com (us.download.nvidia.com)|192.229.232.112|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 443914120 (423M) [application/octet-stream]
Saving to: ‘/dev/null’

/dev/null 100%[===================>] 423.35M 49.4MB/s in 11s

2017-06-25 15:52:56 (39.8 MB/s) - ‘/dev/null’ saved [443914120/443914120]

Kodiack

709 posts

Ultimate Geek

Trusted

#1806480 25-Jun-2017 16:53

gaddman:

I can't remember if you logged a fault already, but assuming you did is it still under investigation or have you been advised that it's fixed? I just tested across a few FibreX connections and got 20-50MBps, eg this FibreX Max in Wellington:

I have indeed logged a fault. Which region is that FibreX Max connection in? I'm in Northland and getting terrible performance to seemingly just about everything but Speedtest.net. Here's what I'm seeing now with a simple wget as well:

--2017-06-25 16:50:51-- http://us.download.nvidia.com/Windows/382.53/382.53-desktop-win10-64bit-international-whql.exe
Resolving us.download.nvidia.com (us.download.nvidia.com)... 2606:2800:247:2063:46e:21d:825:102e, 192.229.232.112
Connecting to us.download.nvidia.com (us.download.nvidia.com)|2606:2800:247:2063:46e:21d:825:102e|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 443914120 (423M) [application/octet-stream]
Saving to: ‘/dev/null’

8% [========> ] 39,505,735 599KB/s eta 15m 47s

These speeds make ADSL look good. :(

Equinox IT

Cloud spending continues to surge globally, but most organisations haven’t made the changes necessary to maximise the value and cost-efficiency benefits of their cloud investments. Download the whitepaper From Overspend to Advantage now.

gaddman

224 posts

Master Geek

Trusted

#1806510 25-Jun-2017 17:17

Kodiack:

I have indeed logged a fault. Which region is that FibreX Max connection in? I'm in Northland and getting terrible performance to seemingly just about everything but Speedtest.net.

That particular one was in Ngaio, but I tested from a few places across Wellington, Kapiti and Christchurch, and they were all in the 10-50MBps range. I notice in your wget you're using IPv6, but so were some of my tests and they were fine. Guessing you've got a physical fault going on, I'm sure the techs will track down a cause soon enough.

kharris

1209 posts

Uber Geek

ID Verified

Trusted

#1806515 25-Jun-2017 17:23

Kodiack:
Needing to update my GPU drivers while troubleshooting an issue. Even this is painfully slow:

I just downloaded my GPU drivers too. Thanks to windows 10 updates. Grrrrr!
Anyway, it was pretty good. 23MB/s. I'm only on a 200Mb connection.
Are you sure you don't have a physical fault?

Kirk

Jase2985

13457 posts

Uber Geek

ID Verified

Lifetime subscriber

#1806517 25-Jun-2017 17:26

MikeHales: No issues on network

it should say "no KNOWN network issues"

the vodafone network has issue, as shown by the few threads on here where people are getting well below advertised speeds, and dropouts etc.

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#1806518 25-Jun-2017 17:28

I'd offer to test v6 over vodafone but looks like my quick router change before leaving has left v6 not being handed out properly..

Much of the v6 routing is not as matured as v4 still.

#include <std_disclaimer>

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

Kodiack

709 posts

Ultimate Geek

Trusted

#1806523 25-Jun-2017 17:48

I don't think it's a physical fault, but I could be wrong. The issues occur both with IPv4 and IPv6. I have disabled IPv6 prior for testing, and performance never did get better. Here's what a speed test looks like at the moment. Certainly much better results than what real-world use is delivering.

gaddman

224 posts

Master Geek

Trusted

#1806530 25-Jun-2017 18:21

@Kodiack, can you provide a traceroute to us.download.nvidia.com ? If it's not a physical fault then maybe it's something with routing back to your IP address.

Kodiack

709 posts

Ultimate Geek

Trusted

#1806535 25-Jun-2017 18:25

In Windows at the moment. Here's how it's looking.

C:\Users\Colton>tracert us.download.nvidia.com

Tracing route to cs5341561.wpc.phicdn.net [2606:2800:247:2063:46e:21d:825:102e]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 2407:7000:9b1a:4e26::1
2 19 ms 8 ms 18 ms 2407:7000::7107
3 11 ms 5 ms 4 ms 2407:7000:f301::1
4 42 ms 39 ms 41 ms 2001:7fa:11:4:0:251c:0:1
5 174 ms 174 ms 191 ms 2001:7fa:11:4:0:12da:0:1
6 167 ms 171 ms 169 ms 2402:7800:0:1::69
7 170 ms 171 ms 171 ms bundle-100.bdr01.syd01.nsw.VOCUS.net.au [2402:7800:0:1::8e]
8 41 ms 39 ms 47 ms as15133.cust.bdr04.syd03.nsw.VOCUS.net.au [2402:7800:0:2::2e6]
9 38 ms 45 ms 40 ms 2606:2800:247:2063:46e:21d:825:102e

Trace complete.

gaddman

224 posts

Master Geek

Trusted

#1806547 25-Jun-2017 19:12

Mmm, looks OK. Let's see what the techs find then, at this point does sound more like a localised physical fault than anything else.

1 | ... | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | ... | 34