Afrihost CAPPED ADSL Feedback (MTN)

[thewusman]

I'll chat to Critical Care about how they arrived at this. At the moment my reports show that we have good headroom on the network to be able to provide excellent throughput. Let me check the exact times from your tests and compare to my reports from the last few nights.

I agree with your comment that the IPC issue is not as bad as it was.

However, I concur with DrJohn that in the evenings, things become less than ideal, to the extent that WA and other accounts (apparently VOX from the results I've seen posted on this forum) outperform WA capped accounts.

That's just what I and a few others seem to be experiencing. Perhaps others are luckier.

 

[Jim West]

Just my 2 cents FWIW, using a 4 Mbps line...

 

[Internet_Junkie]

We'll need to see if you're getting significantly different throughput or latency. A good start would be to run Bitmeter OS during both tests so that we can see if your line is wanting to burst at higher rates and is possibly not able to. The differences will tell us quite a lot.

Watching the same YouTube video

AH

WA

 

[thewusman]

Watching the same YouTube video

AH

WA

So if I read this correctly, much higher bursts on WA, giving a better overall average and sustained throughput ?

 

[Remydp]

Hi everyone,

Does anyone here play BF4 on PS4 by any chance? I've found that with a WebAfrica capped account I consistently get 4/5 bars ping to European servers and with Afrihost I normally get 3/5 and on the odd occasion 4/5. Is there a reason that there is a difference?

 

[DrJohnZoidberg]

Just my 2 cents FWIW, using a 4 Mbps line...

You should run the tests with a bigger sample. 1.3MB and 1.8MB samples are too small and will give you inconsistent results. For a 4Mit line I would try the 6MB download test.

 

[Jim West]

You should run the tests with a bigger sample. 1.3MB and 1.8MB samples are too small and will give you inconsistent results. For a 4Mit line I would try the 6MB download test.

John you are correct. At some point I'll test with larger samples. To be fair any testing should be done systematically as well, not just once-off.

 

[AfriMan]

I agree with your comment that the IPC issue is not as bad as it was.

However, I concur with DrJohn that in the evenings, things become less than ideal, to the extent that WA and other accounts (apparently VOX from the results I've seen posted on this forum) outperform WA capped accounts.

That's just what I and a few others seem to be experiencing. Perhaps others are luckier.

We definitely to go deeper into these issues. Where are we measuring traffic from? If it's international traffic, are we testing to the US, or the EU. All of this gives us a much better idea of where to look. From what I can see we're not breaking our 'sweet spot' for best performance where we would expect to see throughput degrade.

 

[AfriMan]

So if I read this correctly, much higher bursts on WA, giving a better overall average and sustained throughput ?

It's very difficult to say with streaming. I think there it's more important to look at your experience. Is the video buffering or stalling? What is the wait period in between buffers so we can match that to the data. Looking at the graphs, the Afrihost account is still pulling a consistent stream of data, so it may just be a different method of pulling traffic. If the needs of the file don't require more, it's hard to say one is better than the other.

If you're wanting to measure straight throughput, you want to try and saturate your line over a set period and look at the average download rate, or measure the average rate and time to download a specific file. If you have confidence that the server and hosting network is consistent, then this is would be more revealing.

 

[AfriMan]

Hi everyone,

Does anyone here play BF4 on PS4 by any chance? I've found that with a WebAfrica capped account I consistently get 4/5 bars ping to European servers and with Afrihost I normally get 3/5 and on the odd occasion 4/5. Is there a reason that there is a difference?

How is this affecting the performance of your gaming? Are you finding more lag making your gaming unpleasant with us.

Our international cables terminate in the EU so generally EU latency is excellent. Maybe post some UK traceroutes so we can have a look at what sort of latency we're talking about.

 

[AfriMan]

John you are correct. At some point I'll test with larger samples. To be fair any testing should be done systematically as well, not just once-off.

It's best to test over a period of time, and with a set sample that is capable of saturating your line for a good minute or two. Best way to measure throughput, in my opinion.

 

[Jim West]

It's best to test over a period of time, and with a set sample that is capable of saturating your line for a good minute or two. Best way to measure throughput, in my opinion.

+1

 

[Aquiva]

Hi everyone,

Does anyone here play BF4 on PS4 by any chance? I've found that with a WebAfrica capped account I consistently get 4/5 bars ping to European servers and with Afrihost I normally get 3/5 and on the odd occasion 4/5. Is there a reason that there is a difference?

Yeah I do. I've found the same with AH and WA. But on my 10Mb line I don't really see a difference except for when the BF4 server ping shows 1 bar (always on AH, but a router reboot usually sorts it out). I get lag on both ISP's and sometimes disconnects from the voice party's. But this seem to be EA and Sony related. The only thing that makes WA better than AH is streaming (by a big margin). Twitch hardly ever buffers on WA where on AH it is unbearable. Youtube isn't a proper test as they have caching servers apparently. Also users showing the detailed test on this forum shows AH experiences more packet loss as well.

 

[AfriMan]

Yeah I do. I've found the same with AH and WA. But on my 10Mb line I don't really see a difference except for when the BF4 server ping shows 1 bar (always on AH, but a router reboot usually sorts it out). I get lag on both ISP's and sometimes disconnects from the voice party's. But this seem to be EA and Sony related. The only thing that makes WA better than AH is streaming (by a big margin). Twitch hardly ever buffers on WA where on AH it is unbearable. Youtube isn't a proper test as they have caching servers apparently. Also users showing the detailed test on this forum shows AH experiences more packet loss as well.

I've asked our network team to run some end to end tests on streaming servers, focussing on US and EU based services. I believe our bandwidth is solid, and our throughput and latency to the EU is really great. But we'll make sure keep digging. The most important thing is our client's experience, after all.

 

[Remydp]

How is this affecting the performance of your gaming? Are you finding more lag making your gaming unpleasant with us.

Our international cables terminate in the EU so generally EU latency is excellent. Maybe post some UK traceroutes so we can have a look at what sort of latency we're talking about.

I find that gaming on the WA account is a little better than on my AH account but I don't think it's really measurable. I'm not entirely sure how to find the IP Address of a PS4 server but I'll try find out how to do it if possible and post results.

Gaming on my Afrihost account is by no means bad at all, I was just curious as to why they would show different pings to the EU servers. Perhaps there is a small difference but still enough ms for the game to give the extra bar?

Edit: I see you can see the IP Address on Battlelog. Will run some traceroutes tonight

Yeah I do. I've found the same with AH and WA. But on my 10Mb line I don't really see a difference except for when the BF4 server ping shows 1 bar (always on AH, but a router reboot usually sorts it out). I get lag on both ISP's and sometimes disconnects from the voice party's. But this seem to be EA and Sony related. The only thing that makes WA better than AH is streaming (by a big margin). Twitch hardly ever buffers on WA where on AH it is unbearable. Youtube isn't a proper test as they have caching servers apparently. Also users showing the detailed test on this forum shows AH experiences more packet loss as well.

I must say streaming on WA was a lot better than on AH a few weeks ago but streaming on AH seems to have improved a lot recently. I do think that my WA account is better for gaming and streaming but the value for money I get from AH can't be touched.

I signed up for the WA capped account because of their 50gig bonus deal and now I'm contemplating just paying for 5gigs a month which I'll only use on the PS4.

I'm very happy at AH though, was just curious to see if this was the norm or just a once off for some reason on my side

 

[thewusman]

This is the ThinkBroadband test to the UK, 4pm today:

AH capped:
http://www.thinkbroadband.com/speedtest/button/140500427848987083108.png

WA home capped pro:
http://www.thinkbroadband.com/speedtest/button/140500408762021468271.png

A mix of single and multi-threaded testing.

While not saturating the line for a long time, surely there must be a reason for the vast difference in result ?

The linearity of the WA graph seems more expected, whereas the spike and sudden drop on the AH graph is very strange. The AH graph looks good in the morning, then degrades to this and worse later in the evening.

Granted this is just one test, and as Afriman has explained there are many factors, yet I find this hard to explain.

 

[AfriMan]

This is the ThinkBroadband test to the UK, 4pm today:

AH capped:
http://www.thinkbroadband.com/speedtest/button/140500427848987083108.png

WA home capped pro:
http://www.thinkbroadband.com/speedtest/button/140500408762021468271.png

A mix of single and multi-threaded testing.

While not saturating the line for a long time, surely there must be a reason for the vast difference in result ?

The linearity of the WA graph seems more expected, whereas the spike and sudden drop on the AH graph is very strange. The AH graph looks good in the morning, then degrades to this and worse later in the evening.

Granted this is just one test, and as Afriman has explained there are many factors, yet I find this hard to explain.

I ran a single file download over a regular 7Mbps line this morning, downloading a file on a server in Amsterdam. I averaged around 6.5Mbps with peaks just shy of 7Mbps. I ran the test consistently and I'm pretty happy that throughput to the EU/UK is pretty good. I did find that the same test to a US server could be a lot better and I'm taking that to MTN for feedback

Again, I can't speak for various broadband tests that use different methodologies, it's hard to know what links are in that signal chain and whether the are factors outside of our control.

 

[thewusman]

I ran a single file download over a regular 7Mbps line this morning, downloading a file on a server in Amsterdam. I averaged around 6.5Mbps with peaks just shy of 7Mbps. I ran the test consistently and I'm pretty happy that throughput to the EU/UK is pretty good. I did find that the same test to a US server could be a lot better and I'm taking that to MTN for feedback

Again, I can't speak for various broadband tests that use different methodologies, it's hard to know what links are in that signal chain and whether the are factors outside of our control.

Thanks Afriman - appreciate the feedback.

 

[Remydp]

Hi Afriman,

Here are the tracerts to the BF servers.

WebAfrica:

C:\Users\Remy>tracert 159.153.34.107

Tracing route to 159.153.34.107 over a maximum of 30 hops

1 <1 ms <1 ms <1 ms dsldevice.Home [192.168.1.1]
2 7 ms 6 ms 7 ms 196-215-207-1.dynamic.isadsl.co.za [196.215.207.1]
3 8 ms 7 ms 7 ms 196.38.73.85
4 8 ms 8 ms 8 ms cdsl2-rba-vl150.ip.isnet.net [196.38.73.9]
5 9 ms 10 ms 10 ms core1b-pkl-te0-0-0-0.ip.isnet.net [196.26.0.62]
6 180 ms 195 ms 180 ms 168.209.246.3
7 189 ms 181 ms 181 ms xe-4-3-2.edge3.London1.Level3.net [195.50.124.33]
8 191 ms 191 ms 190 ms vl-3503-ve-117.csw1.London1.Level3.net [4.69.166.137]
9 192 ms 192 ms 201 ms ae-58-113.ebr1.London1.Level3.net [4.69.153.121]
10 193 ms 192 ms 192 ms ae-44-44.ebr1.London15.Level3.net [4.69.167.29]
11 197 ms 197 ms 197 ms ae-101-101.ebr2.London15.Level3.net [4.69.167.2]
12 192 ms 192 ms 193 ms ae-5-5.car1.Dublin1.Level3.net [4.69.136.89]
13 192 ms 192 ms 193 ms EA-SWISS-SA.car1.Dublin.Level3.net [213.242.106.
234]
14 191 ms 192 ms 204 ms 159.153.72.22
15 193 ms 193 ms 193 ms 159.153.34.107


And Afrihost:

C:\Users\Remy>tracert 159.153.34.107

Tracing route to 159.153.34.107 over a maximum of 30 hops

1 <1 ms <1 ms <1 ms dsldevice.Home [192.168.1.1]
2 7 ms 6 ms 6 ms 105-236-12-193-esr-lo.mtnbusiness.co.za [105.236.12.193]
3 8 ms 8 ms 8 ms 41.181.221.246
4 35 ms 7 ms 7 ms qux-jh-dca-2.za-b.za.mtnbusiness.net [41.181.165.115]
5 8 ms 7 ms 7 ms jh-dca-2.za--qux-q.za.mtnbusiness.net [196.31.180.4]
6 195 ms 195 ms 195 ms 41.181.180.10
7 195 ms 223 ms 195 ms ls-cr-2.uk--rb-cr-1.za.mtnns.net [196.44.31.114]
8 195 ms 195 ms 282 ms ls-cr-1.uk--ls-cr-2.uk-a.mtnns.net [196.44.31.156]
9 185 ms 184 ms 184 ms ls-pr-2.uk--ls-cr-1.uk-a.mtn.net [209.212.111.203]
10 185 ms 185 ms 185 ms 195.50.122.181
11 207 ms 207 ms 207 ms vl-3506-ve-120.csw1.London1.Level3.net [4.69.166.6]
12 196 ms 196 ms 196 ms ae-57-112.ebr1.London1.Level3.net [4.69.153.117]
13 198 ms 196 ms 199 ms ae-41-41.ebr1.London15.Level3.net [4.69.167.17]
14 207 ms 207 ms 207 ms ae-101-101.ebr2.London15.Level3.net [4.69.167.2]
15 281 ms 256 ms 234 ms ae-5-5.car1.Dublin1.Level3.net [4.69.136.89]
16 197 ms 197 ms 197 ms EA-SWISS-SA.car1.Dublin.Level3.net [213.242.106.234]
17 198 ms 197 ms 198 ms 159.153.72.22
18 197 ms 196 ms 196 ms 159.153.34.107

Trace complete.

I'm not that clued up but looks like they're very similar other than a few spikes in places?

 

[AfriMan]

Hi Afriman,

Here are the tracerts to the BF servers.

WebAfrica:

C:\Users\Remy>tracert 159.153.34.107

Tracing route to 159.153.34.107 over a maximum of 30 hops

1 <1 ms <1 ms <1 ms dsldevice.Home [192.168.1.1]
2 7 ms 6 ms 7 ms 196-215-207-1.dynamic.isadsl.co.za [196.215.207.1]
3 8 ms 7 ms 7 ms 196.38.73.85
4 8 ms 8 ms 8 ms cdsl2-rba-vl150.ip.isnet.net [196.38.73.9]
5 9 ms 10 ms 10 ms core1b-pkl-te0-0-0-0.ip.isnet.net [196.26.0.62]
6 180 ms 195 ms 180 ms 168.209.246.3
7 189 ms 181 ms 181 ms xe-4-3-2.edge3.London1.Level3.net [195.50.124.33]
8 191 ms 191 ms 190 ms vl-3503-ve-117.csw1.London1.Level3.net [4.69.166.137]
9 192 ms 192 ms 201 ms ae-58-113.ebr1.London1.Level3.net [4.69.153.121]
10 193 ms 192 ms 192 ms ae-44-44.ebr1.London15.Level3.net [4.69.167.29]
11 197 ms 197 ms 197 ms ae-101-101.ebr2.London15.Level3.net [4.69.167.2]
12 192 ms 192 ms 193 ms ae-5-5.car1.Dublin1.Level3.net [4.69.136.89]
13 192 ms 192 ms 193 ms EA-SWISS-SA.car1.Dublin.Level3.net [213.242.106.
234]
14 191 ms 192 ms 204 ms 159.153.72.22
15 193 ms 193 ms 193 ms 159.153.34.107


And Afrihost:

C:\Users\Remy>tracert 159.153.34.107

Tracing route to 159.153.34.107 over a maximum of 30 hops

1 <1 ms <1 ms <1 ms dsldevice.Home [192.168.1.1]
2 7 ms 6 ms 6 ms 105-236-12-193-esr-lo.mtnbusiness.co.za [105.236.12.193]
3 8 ms 8 ms 8 ms 41.181.221.246
4 35 ms 7 ms 7 ms qux-jh-dca-2.za-b.za.mtnbusiness.net [41.181.165.115]
5 8 ms 7 ms 7 ms jh-dca-2.za--qux-q.za.mtnbusiness.net [196.31.180.4]
6 195 ms 195 ms 195 ms 41.181.180.10
7 195 ms 223 ms 195 ms ls-cr-2.uk--rb-cr-1.za.mtnns.net [196.44.31.114]
8 195 ms 195 ms 282 ms ls-cr-1.uk--ls-cr-2.uk-a.mtnns.net [196.44.31.156]
9 185 ms 184 ms 184 ms ls-pr-2.uk--ls-cr-1.uk-a.mtn.net [209.212.111.203]
10 185 ms 185 ms 185 ms 195.50.122.181
11 207 ms 207 ms 207 ms vl-3506-ve-120.csw1.London1.Level3.net [4.69.166.6]
12 196 ms 196 ms 196 ms ae-57-112.ebr1.London1.Level3.net [4.69.153.117]
13 198 ms 196 ms 199 ms ae-41-41.ebr1.London15.Level3.net [4.69.167.17]
14 207 ms 207 ms 207 ms ae-101-101.ebr2.London15.Level3.net [4.69.167.2]
15 281 ms 256 ms 234 ms ae-5-5.car1.Dublin1.Level3.net [4.69.136.89]
16 197 ms 197 ms 197 ms EA-SWISS-SA.car1.Dublin.Level3.net [213.242.106.234]
17 198 ms 197 ms 198 ms 159.153.72.22
18 197 ms 196 ms 196 ms 159.153.34.107

Trace complete.

I'm not that clued up but looks like they're very similar other than a few spikes in places?

Yeah, I'd say they're very similar. I'm a little concerned that there is a little more variance in ours, though the end destination latency is the most important one. But if I look at this, the variance and spiking seems to be after you leave our network and go onto Level 3 (which we both peer with) network. So I dunno if this is something that's happening on our bandwidth necessarily.