Afrihost Uncapped ADSL Feedback (Pt3)

Status
Not open for further replies.
W@P said:
I changed from 20mb uncapped to the 200gb capped package to see if things go better. Am i right in saying that i am still on the uncapped till month end ? I changed before the 25th.
Click to expand...

Yes, you should see the package change at month end, unless you moved to a higher value package and pay a pro-rata rate.
 
AfriMan said:
We tested when there was contention on the network. I don't think we proved that gaming is classified as P2P :(
Click to expand...

Nope, there were two separate tests.
One was the latency spikes, which was a contention thing.

The other was the fact that when I was shaped on p2p, I could not even log into diablo while nntp was running at 2.5% of line speed. If I stopped NNTP, login was fine. Started NNTP, lag went through the roof. So something Diablo does was being affected by 25KB/sec NNTP speeds. Only way that can happen on an otherwise unused 10mb line is if it's contending for the same chunk of bandwidth as NNTP - ie, it's in the same bucket. If it was classed as realtime, it would not fall under the 2.5% limit I was seeing.
 
Sinbad said:
You are though. I've proven to you that D3 traffic falls into the same bucket as P2P.
Click to expand...

Wildstar also, only when grouping/adventures and dungeons the traffic is p2p generated as it has to sync your sessions in the group to ensure you are on the correct instance and when this happens your entire session gets shaped untill you disconnect and it releases your IP/packets and reconnect, we are 3 people that stay together and we all 3 have to leave the game and reconnect to remove the lag. 1 person can be in a group and the entire flat gets shaped on Wildstar.

But it was made clear in a nice manner that there is nothing that can be done, the day that afrihost gave gamers a big finger and told them to bad so sad....
 
Sinbad said:
Nope, there were two separate tests.
One was the latency spikes, which was a contention thing.

The other was the fact that when I was shaped on p2p, I could not even log into diablo while nntp was running at 2.5% of line speed. If I stopped NNTP, login was fine. Started NNTP, lag went through the roof. So something Diablo does was being affected by 25KB/sec NNTP speeds. Only way that can happen on an otherwise unused 10mb line is if it's contending for the same chunk of bandwidth as NNTP - ie, it's in the same bucket. If it was classed as realtime, it would not fall under the 2.5% limit I was seeing.
Click to expand...

That sounds more like QoS than being shaped. If traffic is normal when you're not running other services and is affected by running shaped services, then it would not necessarily indicate that the initial service is shaped. I think if we ran a test with Diablo showing that throughput does not exceed the expected percentage on shaped services with nothing else running (while there is no known contention on the network) this would be more definitive.
 
AfriMan said:
That sounds more like QoS than being shaped. If traffic is normal when you're not running other services and is affected by running shaped services, then it would not necessarily indicate that the initial service is shaped. I think if we ran a test with Diablo showing that throughput does not exceed the expected percentage on shaped services with nothing else running (while there is no known contention on the network) this would be more definitive.
Click to expand...

D3 doesn't seem to need more than 25KB/sec though, so I can't prove that one unfortunately.

If p2p streams are being shaped to a very low number, realtime services should still work at best possible speed, correct? Even if you are maxing out the capacity allocated to P2p? Or does the nature of the traffic on your line result in all your traffic being limited to that speed? (ie, he's downloading stuff, so restrict his account to 25KB/sec)
 
Sinbad said:
D3 doesn't seem to need more than 25KB/sec though, so I can't prove that one unfortunately.

If p2p streams are being shaped to a very low number, realtime services should still work at best possible speed, correct? Even if you are maxing out the capacity allocated to P2p? Or does the nature of the traffic on your line result in all your traffic being limited to that speed? (ie, he's downloading stuff, so restrict his account to 25KB/sec)
Click to expand...

as far as I know, it does packet filtering/shaping so its does not shape traffic that is not releated to that packet of data, if the packet has p2p data and for instance wildstar date the entire packet is shaped.

but lets say I run Teamspeak ,dota 2 and a twitch stream all 3 is non p2p/nntp traffic nothing will beshaped.

but if I run Teamspeak, dota 2, and lets say for instance a torrent, then the filter will capture the p2p packet and only shape that and not the packets relating to the other content.
 
B1G_WH1T3 said:
as far as I know, it does packet filtering/shaping so its does not shape traffic that is not releated to that packet of data, if the packet has p2p data and for instance wildstar date the entire packet is shaped.

but lets say I run Teamspeak ,dota 2 and a twitch stream all 3 is non p2p/nntp traffic nothing will beshaped.

but if I run Teamspeak, dota 2, and lets say for instance a torrent, then the filter will capture the p2p packet and only shape that and not the packets relating to the other content.
Click to expand...

Which doesn't explain then why D3 doesn't work while p2p is running at max shaped capacity, unless D3 is falling into the same category.
 
Sinbad said:
Which doesn't explain then why D3 doesn't work while p2p is running at max shaped capacity, unless D3 is falling into the same category.
Click to expand...

Hows your upload while p2p(ing) :D

Make sure you limit upload completely.

Z
 
AfriMan said:
We're definitely not shaping any gaming traffic that is not using P2P traffic. There are some that do, but I don't believe that Diablo is one of them. When you say you try any other account - do you mean a capped, unshaped account from Afrihost?
Click to expand...

I have tried my capped Afrohost account as well as my capped WebAfrica account. My WebAfrica account runs perfectly with 185-195ms in-game latency. My capped Afrihost account however, stays around 270-300ms and spikes to 400+ every few mins. My Afrihost Uncapped 1mb account is unplayable when @ 12.5% shaping. Even @ 25% shaping it gets a bit spikey at times.

Myself and sinbad have provided a fair amount of proof that Diablo 3 is in fact being negatively impacted my p2p shaping... For me its much worse due to my uncapped account only being 1mb. Something is not working as intended on your network but all we get from you guys is that we're wrong...
 
NightowlZA said:
I have tried my capped Afrohost account as well as my capped WebAfrica account. My WebAfrica account runs perfectly with 185-195ms in-game latency. My capped Afrihost account however, stays around 270-300ms and spikes to 400+ every few mins. My Afrihost Uncapped 1mb account is unplayable when @ 12.5% shaping. Even @ 25% shaping it gets a bit spikey at times.

Myself and sinbad have provided a fair amount of proof that Diablo 3 is in fact being negatively impacted my p2p shaping... For me its much worse due to my uncapped account only being 1mb. Something is not working as intended on your network but all we get from you guys is that we're wrong...
Click to expand...

Seems like afrihost need to revisit there way of shaping, changing there system for games, more and more games start to use p2p for instance syncing this is giving a huge impact on the gaming community.

This is why my past 4 years I've preferred webafrica for games, but afrihost got there **** sorted and was a good alternative but now with uncapped being shaped on gaming protocols looks like it jumping back time to webafrica uncapped.
 
NightowlZA said:
Myself and sinbad have provided a fair amount of proof that Diablo 3 is in fact being negatively impacted my p2p shaping... For me its much worse due to my uncapped account only being 1mb. Something is not working as intended on your network but all we get from you guys is that we're wrong...
Click to expand...

+1 for this! I've been visiting this forum the past few weeks, and I've seen numerous people complain about D3 and Wildstar's pings being affected by shaping. I don't know why Afriman keeps on denying this, but surely its worth chatting to the technicians about. But instead, they give the 'you're wrong'. Very poor support service. What ever happened to 'the customer is always right'?
 
To be fair in most tech fields the customer is normally clueless;-)

I'm not though. This is my vocation.
 
Sinbad said:
To be fair in most tech fields the customer is normally clueless;-)

I'm not though. This is my vocation.
Click to expand...

Agreed and +1 :)

To be fair, this is myBB where we all discuss things in fairly low level and technical detail...

The reps could easily make some inferences from this, IE: that we're all in fact not a bunch of technically-clueless rocks :P
 
I think the reps here are generally first or second line support. No disrespect to them meant, but they kinda have to go on what their 3rd line and admin /architects say. I doubt that someone employed at say ccie level would have time to do client facing work on a forum.
 
Sinbad said:
I think the reps here are generally first or second line support. No disrespect to them meant, but they kinda have to go on what their 3rd line and admin /architects say. I doubt that someone employed at say ccie level would have time to do client facing work on a forum.
Click to expand...

I won't claim to be a systems engineer, but I do work with a very transparent team. I'm pretty sure that the information I have is verified and correct. I also think from the perspective of running networks the size of Afrihost and MTN are very different in terms of scale and application, so it's not always a direct translation to experience on smaller networks.
 
Last edited:
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X