Afrihost Uncapped ADSL Feedback (Pt3)

Status
Not open for further replies.
Not strictly true. Especially if you're measuring response times of large packets in quantity (eg, D3 traffic)....

I think we've gone over the D3 stuff, our team should have been in touch with you around the details?
 
I think we've gone over the D3 stuff, our team should have been in touch with you around the details?

No contact whatsoever, but irrelevant as I am no longer an uncapped customer.
 
No contact whatsoever, but irrelevant as I am no longer an uncapped customer.

I'm sure testing was done on this, from the conversations I remember we weren't able to replicate the issues.
Shout if you want a call and another investigation done from our team :)
 
I'm sure testing was done on this, from the conversations I remember we weren't able to replicate the issues.
Shout if you want a call and another investigation done from our team :)
If I ever come back (unlikely given the new policy of not being allowed to do legitimate downloads during the evening), I will do so.
 
You have no idea what you are talking about

Afrihost has always shaped me lightly because I don't use much data. i know about the general shaping rule blada blada etc but shaping is always ok for me.

Thanks for the feedback, I'm glad that the products works well for your purposes. :)

My pleasure :)

I am a firm believer in that everybody can voice their opinion.
So here goes.

How does he get minimal shaping on his account?

.
Do you agree that is an inaccurate statement?
Do you agree that in fact AH shapes all downloads to its fullest from 06:00 - 22:00 and that 12.5% and 6.25% leaves you with very little line speed for downloads during that time period and that dynamic shaping from 22:00 rarely lifts before 02:00 or do you classify that as light shaping?

Lets talk figures.

My 4Mbps line = 3400kbps on a speedtest.
3400kbps /8 = 425kB/s. That is what I should get on downloads.

Now apply AH light shaping and you get :
425kB/s * 12.5% light shaping from 06:00 - 22:00>

53.1kB/s for Http downloads and about 26.5kB/s for P2P

Now please explain to me how do you see that as light shaping?

I have a 1Mb line so shaping doesn't seem to slow me down that much to be honest. HTTP downloads flies all the time. I schedule the rest of my downloads during the night/early hours of the mornings.

Wouldn't gifs and jpegs be counted as realtime? This page has about fifteen jpegs that I can see from the source code. Also, to Mr http download is perfect for me, what speeds are you getting on what speed line, have you ever used an ISP besides afrihost and were/are you admitted to a phsychiatric hospital?

LOL @ the psych hospital part.

I've used Telkom, Mweb, Openweb and Plugg.
 
Afrihost has always shaped me lightly because I don't use much data. i know about the general shaping rule blada blada etc but shaping is always ok for me.


Please take a screenshot of your shaping % please when you get a chance.

Since you are on 1mb...12.5% of that should equate to around 12kB/s odd or there about.

If that is what you consider flying...then no.

If you are indeed somehow exempt from the mandatory shaping during the day I'd love to see it.
 
Something that also bugs me is this: using RoMark's calculations the speeds I'm supposed to get while shaped is as follows:
My 2Mbps line = 1700kbps on speedtest
1700kbps /8 = 212kB/s which is max theoretical speeds depending on source
212kB/s * 25% shaping as per Afrihost clientzone should = 53kB/s
but I don't get even that, my average, on all HTTP & p2p is a paltry 5kB/s

I think we're being lied to, Afriman please explain

Edit: as per clientzone I'm shaped 6.5% for p2p, which should = 14kB/s (rounded off)
 
Last edited:
Shaping doesn't affect latency in any way :)

Unless if you put ICMP on a preemptive shaper, there will be a latency increase of the buffer and whatever other queues came in before the shaper tended to the ICMP packet. If you look at the latency of transactional data end-end, shaping does affect latency and jitter, no two ways about it. If you drop packets (the feedback of a shaper telling the sender to ease off), the round trip is at very least that of the next hop's latency + a random delay, which for international can be a fair bit.

Your shaper (or whatever other connections upstream) will drop packets on inbound TCP packets to regulate the speed. That introduces latencies and jitter at very best of a good few packets at the outbound packet rate + more. Just because a ping passed through the shaper with little latency doesn't imply that that's the case for other traffic.
 
Unless if you put ICMP on a preemptive shaper, there will be a latency increase of the buffer and whatever other queues came in before the shaper tended to the ICMP packet. If you look at the latency of transactional data end-end, shaping does affect latency and jitter, no two ways about it. If you drop packets (the feedback of a shaper telling the sender to ease off), the round trip is at very least that of the next hop's latency + a random delay, which for international can be a fair bit.

Your shaper (or whatever other connections upstream) will drop packets on inbound TCP packets to regulate the speed. That introduces latencies and jitter at very best of a good few packets at the outbound packet rate + more. Just because a ping passed through the shaper with little latency doesn't imply that that's the case for other traffic.

It may affect is by a few ms, but not anything noticeable, whether the account is shaped/ unshaped - the latency stays the same and won't affect your experience :)
 
Please take a screenshot of your shaping % please when you get a chance.

Since you are on 1mb...12.5% of that should equate to around 12kB/s odd or there about.

If that is what you consider flying...then no.

If you are indeed somehow exempt from the mandatory shaping during the day I'd love to see it.


I get around 25% at the very strictest level for http, and 12.5% for P2P. Never been below this so far yet... (holds humbs)
 
It may affect is by a few ms, but not anything noticeable, whether the account is shaped/ unshaped - the latency stays the same and won't affect your experience :)

I can only hope that your network's inability to compete with IS on international content is not as a result of such an attitude behind the scenes.

Jitter of 20ms on a 190ms with 2% packet loss is very noticeable on gaming for example. When I tested a VPN which added ~25ms latency but had essentially zero packet loss and very little jitter the experience improved notably.

There are some auto-admin servers that will kick you when the latency degrades. It's the difference between carrying on with a session or getting kicked.
 
I can only hope that your network's inability to compete with IS on international content is not as a result of such an attitude behind the scenes.

Jitter of 20ms on a 190ms with 2% packet loss is very noticeable on gaming for example. When I tested a VPN which added ~25ms latency but had essentially zero packet loss and very little jitter the experience improved notably.

There are some auto-admin servers that will kick you when the latency degrades. It's the difference between carrying on with a session or getting kicked.

I've got full confidence in our network, from my first hand experience I've seen nothing short of awesome.
Regardless, I don't think either you or I can speculating around how, if or how much latency affects shaping on a network this size ;)
 
Unfortunately my experience differs. Using a locally hosted VPN for international has sorted all my issues, effectively isolating the quality of your international back haul and motivating me to move off AH. I am still seeing ~1mbps throughput to German video servers on a 10 meg line (capped, I know this is the uncapped thread but I was using uncapped till recently). Changing ISPs to an IS reseller sees video buffering at the line speed and if not, there isn't jerky speed throughput as with AH. IS is definitely doing something far better than MTN/Afrihost in this space.
 
Unfortunately my experience differs. Using a locally hosted VPN for international has sorted all my issues, effectively isolating the quality of your international back haul and motivating me to move off AH. I am still seeing ~1mbps throughput to German video servers on a 10 meg line (capped, I know this is the uncapped thread but I was using uncapped till recently). Changing ISPs to an IS reseller sees video buffering at the line speed and if not, there isn't jerky speed throughput as with AH. IS is definitely doing something far better than MTN/Afrihost in this space.

I think this needs to be taken to a platform where more detail can be discussed - have you submitted a ticket to our team? Higher level feedback always gets seen by our network team :)
 
I get around 25% at the very strictest level for http, and 12.5% for P2P. Never been below this so far yet... (holds humbs)


I am not a Home uncapped customer any more, but did Afriman not say that all Home uncapped users are shaped to max?

In fact this is his answer on my post:
We shape from 7:30 to 10pm at a set value for all Uncapped users, thereafter we shape dynamically.

So you should also be shaped to 12.5 for http and 6.25 for P2P.
I might be wrong, as I said moved from this account a while back.

P.S
Using my calculations, I think you should theoretically get about 110kB/s on a 1 Mbps line? Might be even a bit less, depending on your line, but 110kB/s * 25% = 27.5kB/s.

So instead of 110kB/s you can only only download at 27.5kB/s on Http.

Not picking a fight, just saying that at at 25% shaping you lose 75% of your possible download speed and at 12.5% you lose 87.5% from possible Http speed and at 6.25% shaping on P2P you lose 93.75 of your possible download speed.

That is not really light shaping if you think about it.
 
Unfortunately my experience differs. Using a locally hosted VPN for international has sorted all my issues, effectively isolating the quality of your international back haul and motivating me to move off AH. I am still seeing ~1mbps throughput to German video servers on a 10 meg line (capped, I know this is the uncapped thread but I was using uncapped till recently). Changing ISPs to an IS reseller sees video buffering at the line speed and if not, there isn't jerky speed throughput as with AH. IS is definitely doing something far better than MTN/Afrihost in this space.

Agreed 300% with your point regarding IS' network quality vs MTN.

I once worked for a company that used MTN as a corporate ISP. MTN Business sold us a 100Mb internet breakout, 1:1 contention with <supposedly> 100% international portion (100Mb).
My international tests at best showed around 3Mb/s.
MTN Business also have no idea how to scale their network.

@Afriguy - some of us have worked with much bigger networks than that of MTN. I believer Timber_MG has made some valid points. Unfortunately I've already moved as well (probably to the same IS reseller as Timber_MG).
 
Agreed 300% with your point regarding IS' network quality vs MTN.

I once worked for a company that used MTN as a corporate ISP. MTN Business sold us a 100Mb internet breakout, 1:1 contention with <supposedly> 100% international portion (100Mb).
My international tests at best showed around 3Mb/s.
MTN Business also have no idea how to scale their network.

@Afriguy - some of us have worked with much bigger networks than that of MTN. I believer Timber_MG has made some valid points. Unfortunately I've already moved as well (probably to the same IS reseller as Timber_MG).

We've been scaling our IPC for years, it goes a bit above and beyond a 100Mb link :)
 
I'm referring to MTN's backbone. That goes far beyond an IPC link :)

Indeed it does, from the monitors we run and scaling we've done no both our local and international links - MTN has been there all the way.
 
Status
Not open for further replies.
Top
Sign up to the MyBroadband newsletter
X