Afrihost Uncapped ADSL Feedback

Status
Not open for further replies.
Did some tests last night on my 4mbp uncapped from PE from 20:00 onwards.

Streamed 1080p HD from Youtube (Iron Maiden live if you must know), and it was seamless with no buffering in full screen mode. Downloaded one of the 350mb videos - came in at 400kb/s.

Did have a strange issue when downloading random files from mirror.ac.za. A couple of the files started downloading at 400kb/s and then rapidly died away until I got a Network error. This only occurred for the slackware mirror, other mirrors downloaded through http at +300kb/s. Didn't try ftp.

Latency - when using Mweb I get 30ms to CT, when using AH I get 51ms. Will only get a true reflection once IPC has been upgraded.
 
new_in_za2 said:
In reality it has nothing at all to do with latency on an operating system released after 1997. You really need to learn about how TCP works. Sorry. See here.
Click to expand...

Hi,

Unfortunately you are slightly mistaken. Latency does have an effect on the way traffic flows.You are correct however that certain ports and sites seem to be prioritized. Here is an example:

To my UK server running iperf as a test here is traffic flowing over a non standard port:

------------------------------------------------------------
Client connecting to 91.xxx.xxx.xxx, TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[224] local 192.168.3.6 port 53019 connected with 91.xxx.xxx.xxx port 5001
[216] local 192.168.3.6 port 53018 connected with 91.xxx.xxx.xxx port 5001
[188] local 192.168.3.6 port 53016 connected with 91.xxx.xxx.xxx port 5001
[220] local 192.168.3.6 port 53017 connected with 91.xxx.xxx.xxx port 5001
[ ID] Interval Transfer Bandwidth
[220] 0.0-13.2 sec 256 KBytes 159 Kbits/sec
[216] 0.0-13.4 sec 256 KBytes 157 Kbits/sec
[188] 0.0-13.4 sec 256 KBytes 156 Kbits/sec
[224] 0.0-13.9 sec 768 KBytes 452 Kbits/sec
[SUM] 0.0-13.9 sec 1.50 MBytes 905 Kbits/sec

C:\Users\Gandalf\Desktop\Iperf\Release>Iperf.exe -c 91.xxx.xxx.xxx
------------------------------------------------------------
Client connecting to 91.xxx.xxx.xxx, TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[156] local 192.168.3.6 port 53021 connected with 91.xxx.xxx.xxx port 5001
[ ID] Interval Transfer Bandwidth
[156] 0.0-10.4 sec 1.00 MBytes 804 Kbits/sec

C:\Users\Gandalf\Desktop\Iperf\Release>

As you can see the multiple streams and single stream in this instance appear similar. This is due to AH not recognizing the traffic and shaping it down (905 vs 804)

However, if i create an encrypted tunnel and do the same test over the tunnel running on 443:

------------------------------------------------------------
Client connecting to 91.xxx.xxx.xxx, TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[224] local 172.16.10.131 port 53257 connected with 91.xxx.xxx.xxx port 5001
[220] local 172.16.10.131 port 53256 connected with 91.xxx.xxx.xxx port 5001
[184] local 172.16.10.131 port 53255 connected with 91.xxx.xxx.xxx port 5001
[208] local 172.16.10.131 port 53254 connected with 91.xxx.xxx.xxx port 5001
[ ID] Interval Transfer Bandwidth
[220] 0.0-11.0 sec 2.88 MBytes 2.20 Mbits/sec
[208] 0.0-11.1 sec 1.50 MBytes 1.13 Mbits/sec
[224] 0.0-11.7 sec 1.00 MBytes 716 Kbits/sec
[184] 0.0-13.1 sec 1.00 MBytes 642 Kbits/sec
[SUM] 0.0-13.1 sec 6.38 MBytes 4.09 Mbits/sec

C:\Users\Gandalf\Desktop\Iperf\Release>Iperf.exe -c 91.xxx.xxx.xxx
------------------------------------------------------------
Client connecting to 91.xxx.xxx.xxx, TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[156] local 172.16.10.131 port 53259 connected with 91.xxx.xxx.xxx port 5001
[ ID] Interval Transfer Bandwidth
[156] 0.0-10.3 sec 4.00 MBytes 3.27 Mbits/sec

The results are fairly typical, I get on average around 1meg more, if I run them several times and collate the results.

The overall speeds are pretty poor, on a 10meg uncapped business account international is 4-5Megs at best. Unknown traffic is lower around 1meg as you can see, and any shaped traffic seems to follow that trend.

Hardly astounding results.

Pete

Edit: Clarification - only testing international traffic here.
 
Last edited:
whats going on there is a green light but it took the better part of 3mins to load the forum, speedtest wont load, international sites just time out... HELP
 
2367657687.png


2367661559.png
 
remus786 said:
Stop using Speedtest, it is by no means a useful tool as AH prioritize the site.
Click to expand...

Hi
1) Please justify this statement. That is a serious allegation - and does not gel with my experience of the network.
2) Which site? Speedtest.net consists of many hundreds of servers around the world.


Clinton which London server did you use? I found the vodafone one to be congested, I tried the Fidonet one.
 
blunt said:
Quiet thread, probably a good reflection on the product :)

For me - Gmail was rather slow yesterday but not un-usable, other than that it's been perfect.
Click to expand...

Or maybe the faith has been lost? Too many issues lately from what I see and hear.
 
remus786 said:
Stop using Speedtest, it is by no means a useful tool as AH prioritize the site.
Click to expand...

Using speedtest, the results coincide between Saix.net, MyBB and Speedtest.net. This also coincides with my http download speed. I am sure if they went to the trouble to prioritise one they would simply add the rest, testmy.net included. One line of config would be enough for all sites.

Urban myth?
 
roadcat said:
Using speedtest, the results coincide between Saix.net, MyBB and Speedtest.net. This also coincides with my http download speed. I am sure if they went to the trouble to prioritise one they would simply add the rest, testmy.net included. One line of config would be enough for all sites.

Urban myth?
Click to expand...

I think more likely a malicious rumour started by people with an agenda against afrihost.
 
Whats happened to Astraweb? Found it to be 15kbs on 2meg uncapped switched to capped, started at 200kbs then dropped to 0 (I suspect it started with news.afrihost.com and when it wanted to get the unknown groups but known to Astraweb ..... )
 
Sinbad said:
I think more likely a malicious rumour started by people with an agenda against afrihost.
Click to expand...

Agree. In my experience AH do try to help if you provide them enough information about your problem. If they were just trying to trick people with false download speeds, they wouldn't have bothered helping as much as they do.
 
International test
Test conducted on 12 December 2012 21:58
Download Speed: 75 kbps (9.4 KB/sec transfer rate)
Upload Speed: 292 kbps (36.5 KB/sec transfer rate)
Latency: 285 ms

Speedtest doesn't even load

Tracing route to eu.battle.net [80.239.186.40]
over a maximum of 30 hops:

1 1 ms <1 ms <1 ms 192.168.0.1
2 42 ms 8 ms 8 ms 105-236-8-65-esr-lo.mtnbusiness.co.za [105.236.8.65]
3 42 ms 30 ms 50 ms ipc-recieve-rb-2a.za.mtnbusiness.net [41.181.178.65]
4 32 ms 35 ms 30 ms rb-dca-2.za--rb-cr-1.za-a.mtnns.net [196.44.31.120]
5 31 ms 31 ms 45 ms jh-cr-1.za--rb-cr-2.za.mtnns.net [196.44.0.42]
6 46 ms 51 ms 99 ms qux-jh-dca-2.za-b.za.mtnbusiness.net [41.181.165.115]
7 248 ms 229 ms 208 ms xe-4-1-0.edge5.Amsterdam1.Level3.net [212.72.41.89]
8 243 ms 224 ms 224 ms 4.69.162.129
9 258 ms 213 ms 213 ms ae-1-51.edge4.Amsterdam1.Level3.net [4.69.139.138]
10 229 ms 259 ms 223 ms Telia-level3-10g.Amsterdam1.level3.net [4.68.62.98]
11 214 ms 208 ms 251 ms adm-bb4-link.telia.net [213.155.133.208]
12 269 ms 283 ms 280 ms prs-bb1-link.telia.net [80.91.245.69]
13 227 ms 228 ms 241 ms prs-b8-link.telia.net [213.155.131.3]
14 * * * Request timed out.
 
Sent Afriman a PM. Getting 888bps from Microsoft downloads, and 2kbps from cdn.mozilla.com on http downloads.
 
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X