Afrihost Uncapped ADSL Feedback

Status
Not open for further replies.
Here's some facts about what is happening. Someone posted in one of the threads that it's unfair to blame heavy users. I can understand that, we should have penalised them more brutally, but we tried to give them a good quality experience by de-prioritising traffic but still allowing them to torrent at reasonable speeds. At the moment torrents and downloads account for 2/3 of bandwidth usage. Majority of this has been between 5pm and midnight. On average our clients moved only 70% of the bandwidth on IS than is being used currently on MTN, while our projected average shows a 43% increase in total bandwidth used over a one month period. That's a lot more bandwidth being supplied to clients than ever before, and nearly 70% of that is being used up in torrents.

While I would personally love to ask the top percentage of users (who alone account for nearly 200TB of the bandwidth used) exactly what they need all these linux distro's for, we are going to target them specifically by limiting their torrent and download capacity during busy hours, but preserve their real time usage at all times.

I know negativity builds momentum faster than positivity. And obviously there are a lot of negative feelings out there at the moment. I do find it interesting that some guys who are posting here have cancelled, been refunded, and yet still are posting complaints about a service they are no longer using. That boggles my mind :)

You do know that most Steam games are above 10 gigs? And with all the specials running one can easily download over 200 gigs in games. Its funny heavy usage is always compared to piracy
 
Last edited:
ISP of the year 2012... Well until the 16th October... Like for Lens Armstrong, I think they should take it back!
 
ISP of the year 2012... Well until the 16th October... Like for Lens Armstrong, I think they should take it back!

Expect them to bounce back.

Form is temporary, class is permanent.

One of life's most important lessons.
 
Thanks to everyone who replied to my PM's, we really appreciate your assistance in getting things right. Now for the apology. With Gian having asked guys to PM and email me directly, I'm busy addressing all the messages :(
I should have replied and confirmed with everyone by the end of day today. So please be patient - ek het net twee hande (as the saying goes). This shouldn't prevent you in any way to keep the results flowing, it's all being compiled for the team so they can be as reactive (and pro-active) as possible.

And thank you also for the feedback which some have emailed directly to us following Gian's statement, we are going to start addressing each of your issues directly.

From the feedback, it seems mixed from last night. Lots of improvement definitely noted, but seems like latency spiked for some around 9pm last night. Also looks like torrents and NZB's were going full ball while HTTP started to stutter, should have been the other way round mostly. We've matched the feedback with changes being adjusted at the time, and this has given the team a few ideas about where the bulk of the issues may be. The team has been here all night, have gone home for a shower & shave and are back in the war room hammering this out. We are going to get this right, we know it's possible and we have the right people to achieve this.
 
Thanks to everyone who replied to my PM's, we really appreciate your assistance in getting things right. Now for the apology. With Gian having asked guys to PM and email me directly, I'm busy addressing all the messages :(
I should have replied and confirmed with everyone by the end of day today. So please be patient - ek het net twee hande (as the saying goes). This shouldn't prevent you in any way to keep the results flowing, it's all being compiled for the team so they can be as reactive (and pro-active) as possible.

And thank you also for the feedback which some have emailed directly to us following Gian's statement, we are going to start addressing each of your issues directly.

From the feedback, it seems mixed from last night. Lots of improvement definitely noted, but seems like latency spiked for some around 9pm last night. Also looks like torrents and NZB's were going full ball while HTTP started to stutter, should have been the other way round mostly. We've matched the feedback with changes being adjusted at the time, and this has given the team a few ideas about where the bulk of the issues may be. The team has been here all night, have gone home for a shower & shave and are back in the war room hammering this out. We are going to get this right, we know it's possible and we have the right people to achieve this.

+1 :)
 
must admit, the international D/L dropped considerably between 0845 and 0900, then stabilized. skype didnt drop but had some awful stutter, otherwise sofar so good.. even latency is constant, which makes us gamers happy..
 
It would appear as if there is some improvement as of yesterday compared to where traffic was throttled to 1 mbps.

Shaping was definitely in operation, though it wan't all that constant (which may be indicative of some packet loss or a very short time-period on their shapers). I got Youtube streams at about 1.5-2.5 mbps, torrents were crawling at around 250 kbps with occasional peaks in the 1 mbps range. A normal single threaded download ran at 300 kB/s and stayed there pretty much when streaming above that.

When I shape my internal users I do it on a per user basis and not on a per connection basis. When the line delivers full speed (thinking back to the good days on IS) multi-threaded downloads with say 20 connections would out-contend anything else 20:1 until I implemented a shaping process that was aware of the destination of the traffic and categorized traffic by port and by deep packet inspection. Then all of a sudden it would contend faily or dynamically de-prioritize different classes.

I wonder whether we'll see burst-type shaping on some protocols like we saw with IS during the day say a month before the change to MTN.
 
Last edited:
Thanks to everyone who replied to my PM's, we really appreciate your assistance in getting things right. Now for the apology. With Gian having asked guys to PM and email me directly, I'm busy addressing all the messages :(
I should have replied and confirmed with everyone by the end of day today. So please be patient - ek het net twee hande (as the saying goes). This shouldn't prevent you in any way to keep the results flowing, it's all being compiled for the team so they can be as reactive (and pro-active) as possible.

And thank you also for the feedback which some have emailed directly to us following Gian's statement, we are going to start addressing each of your issues directly.

From the feedback, it seems mixed from last night. Lots of improvement definitely noted, but seems like latency spiked for some around 9pm last night. Also looks like torrents and NZB's were going full ball while HTTP started to stutter, should have been the other way round mostly. We've matched the feedback with changes being adjusted at the time, and this has given the team a few ideas about where the bulk of the issues may be. The team has been here all night, have gone home for a shower & shave and are back in the war room hammering this out. We are going to get this right, we know it's possible and we have the right people to achieve this.

+10 :)
 
Last edited:
My line has been stable since yesterday at 3PM

2293705955.png
 
My current speedtest to

International

Test conducted on 11/8/2012 9:44:39 AM

Download Speed: 127 kbps (15.9 KB/sec transfer rate)
Upload Speed: 304 kbps (38 KB/sec transfer rate)
Latency: 252 ms

Local

Test conducted on 11/8/2012 9:55:16 AM

Download Speed: 313 kbps (39.1 KB/sec transfer rate)
Upload Speed: 400 kbps (50 KB/sec transfer rate)
Latency: 103 ms

2Mb uncapped
 
http is dead for me
nntp / torrents is fine.

supposed to be the other way round
 
My current speedtest to

International

Test conducted on 11/8/2012 9:44:39 AM

Download Speed: 127 kbps (15.9 KB/sec transfer rate)
Upload Speed: 304 kbps (38 KB/sec transfer rate)
Latency: 252 ms

Local

Test conducted on 11/8/2012 9:55:16 AM

Download Speed: 313 kbps (39.1 KB/sec transfer rate)
Upload Speed: 400 kbps (50 KB/sec transfer rate)
Latency: 103 ms

2Mb uncapped

KoKaRoT, seeing as I'm in the same boat as you (2 Mbps, bad speeds, etc.), I'd be interested to see what happens if you use the Telkom test account (username: guest@telkomadsl; password: guest) and do a speedtest on http://saix.net. Are the speeds as bad as with your Afrihost account?
 
Thanks to everyone who replied to my PM's, we really appreciate your assistance in getting things right. Now for the apology. With Gian having asked guys to PM and email me directly, I'm busy addressing all the messages :(.

No problem.

Things from my side have seemed to improve but I had a bad Skype session last night. (10-11pm) Very laggy and had 3 disconnects. Never had problems before. It could just be a issue with Skype - there were 43 million people online! :eek:
 
My 35 gig account was not usable this morning with http. I have swapped over to a saix account until this is sorted out.
 
Status
Not open for further replies.
Top
Sign up to the MyBroadband newsletter
X