Afrihost Business Uncapped Feedback

[Simple.]

Same issue regarding international speeds and pings. Ping to US is ~80ms higher than normal.
Max HTTP and streaming speeds are around 200Kbps on my 4Mbps line.
Been like this for the past week. Have the same issue at work too.
Capped accounts and WebAfrica run at full line speed.

Any word from Afrihost on this issue?

 

[Rohan]

Are you in the southern part of the country? Capped is not working properly at all in the Cape this evening (about 20% of line speed), so I assume that if capped is screwed, uncapped will be even worse.

Nope I am in North West. Speeds are now back to normal - after a whole afternoon and evening at 1m.

 

[Hummercellc]

I have noticed packet loss on all international traffic. Local is fine. This is making Skype unusable...

 

[JesseG]

I have no other issue with AH besides the spike in latency on the internal network, last Saturday I was streaming twitch on high quality without any buffering or drop in quality for 2 hours straight, my line is near maxing on full speed when downloading, Youtube is streaming OK so far...Please fix this problem and I will be a happy camper * on COD* lol

 

[AfriMan]

http://imgur.com/Y34t44p
This is a screenshot I sent your customer care team 3 months ago (as can be seen by the upload date on Imgur); MTN has been having issues even before that and you're telling us this isn't confirmed yet? What more proof do you need when a network has an internal local spike of 200ms?

The months before Afrihost joined MTN's network were glorious but since then the service has been appalling. This is one of the reasons I dropped them as my provider.

The spike looks like it's the hop from our network to UK network, basically having travelled the length of the undersea cable. That would be pretty normal. I can't see the description but a jump in latency between the local network and Amsterdam is pretty normal.

 

[AfriMan]

View attachment 149756

I'm sending this to the feedback e-mail addy as well, but damn. Youtube buffers on 480p and I'm the only one doing anything on the network at home. Is this related to the two issues causing a medium impact on the network?

Jeffries Bay would be part of the South and we are seeing some diminished throughput due to increased demand. We're waiting for some software upgrades to kick in that will over the next few days that will manage demand more efficiently and hopefully resolve this

 

[AfriMan]

Things are working fine for me. Getting just below advertised speeds, no lag spikes in games and streaming is working like a charm. *Touches wood*

EDIT: I still wonder, though, why with Afrihost on MTN's "fast fibre network" that the pings are higher than that of Telkom. Really bums me out -- literally the only thing Telkom has an advantage on.

In theory, Telkom's pings should always be better because they don't need to break out of their network, but I am not 100% how their internal network works, because they do use IPC. So you have to travel over the copper infrastructure in your area, via the exchange and over Telkom's network until you reach our closest IPC before you enter MTN's fibre network. So it really depends on how long (and what state of repair) the previous steps are on Telkom's network, and how that routing differs between how they route their traffic and ours (becuase it depends on where their infrastructure is).

 

[AfriMan]

Still getting 1m speeds on international on my 10m line. Web Africa is running full speed. Not getting any real answers from critical care except there are issues on the MTN network but they need to prove it to MTN? Local speeds are not affected at all. I really hope you are not starting to shape the business accounts like Axxess is now doing.

I think we're running a lot of tests with MTN to try to determine exactly where international speeds are poor and how they can be improved. From my last feedback, our UK/EU traffic is excellent but US traffic could be better and we're escalating to our US partners via MTN for them to improve wherever possible

 

[AfriMan]

Are you in the southern part of the country? Capped is not working properly at all in the Cape this evening (about 20% of line speed), so I assume that if capped is screwed, uncapped will be even worse.

The issue is that we're seeing increased demand over a period between 7pm and 10pm, and this is causing contention over the South network. This would affect throughput on all traffic in the South. We're updating and optimising the network management software to better manage overall traffic management, and we are confident that this will relieve the contention we're seeing. But we'll keep working on this until we find the appropriate solution to address the experience clients are getting

 

[AfriMan]

I have noticed packet loss on all international traffic. Local is fine. This is making Skype unusable...

We're seeing diminished throughput, but latency and packet loss should be normal. Have you run an MTR tests so I can see where the packet loss is occurring?

 

[JesseG]

The spike looks like it's the hop from our network to UK network, basically having travelled the length of the undersea cable. That would be pretty normal. I can't see the description but a jump in latency between the local network and Amsterdam is pretty normal.

Normal??? If I run a tracert from another ISP there is no internal spike in latency, the only spike is when it switches from the ISP network to UK network...I don't think it's normal, if I run the tracert from your network then the spike starts at hop 5 while still on the MTN servers, it's not normal...So how are other ISP's able to eliminate internal spikes if this is normal? I don't understand the whole technical aspect of it all, just a few bits here and there but it doesn't seem like a normal thing to me...

 

[AfriMan]

Normal??? If I run a tracert from another ISP there is no internal spike in latency, the only spike is when it switches from the ISP network to UK network...I don't think it's normal, if I run the tracert from your network then the spike starts at hop 5 while still on the MTN servers, it's not normal...So how are other ISP's able to eliminate internal spikes if this is normal? I don't understand the whole technical aspect of it all, just a few bits here and there but it doesn't seem like a normal thing to me...

It doesn't really matter where the latency spikes, but we can expect it to spike where it leaves the local network, goes over the international cable and then comes back for the ping time. If that is then consistent with the overall pattern of ping and the end ping is within acceptable range, then I think that's pretty normal.

 

[Ho3n3r]

http://imgur.com/Y34t44p
This is a screenshot I sent your customer care team 3 months ago (as can be seen by the upload date on Imgur); MTN has been having issues even before that and you're telling us this isn't confirmed yet? What more proof do you need when a network has an internal local spike of 200ms?

The months before Afrihost joined MTN's network were glorious but since then the service has been appalling. This is one of the reasons I dropped them as my provider.

Uhhh... yeah that looks pretty normal to me. Not sure what your complaint is? It says SA, but it is cleary in Europe, since the 2nd last to last ping/hop is only 5ms. Logic dictates that NO data will travel from SA to Europe in 5 milliseconds.

 

[JesseG]

Here is tracert on Afrihost network :
C:\Users\jesseg>tracert www.tweaknews.eu

Tracing route to www.tweaknews.eu [82.197.196.59]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 10.0.0.2
2 8 ms 7 ms 7 ms 105-236-0-65-esr-lo.mtnbusiness.co.za [105.236.0
.65]
3 17 ms 17 ms 28 ms 41.181.221.218
4 22 ms 16 ms 18 ms rb-dca-2.za--qux-b.za.mtnbusiness.net [41.181.16
5.118]
5 208 ms 208 ms 207 ms xp-dca-1.za--xp-cr-1.za-a.mtnns.net [196.44.0.15
5]
6 208 ms 208 ms 208 ms jh-cr-1.za--rb-cr-2.za.mtnns.net [196.44.0.42]
7 230 ms 215 ms 215 ms ls-cr-2.uk--rb-cr-1.za-g.mtnns.net [196.44.31.11
4]
8 217 ms 217 ms 230 ms am-cr-1.nl--ls-cr-2.uk-a.mtnns.net [196.44.31.18
3]
9 224 ms 223 ms 224 ms am-tpr-1.nl--am-cr-1.nl-a.mtn.net [209.212.111.1
41]
10 223 ms 223 ms 233 ms xe-4-1-0.edge5.Amsterdam1.Level3.net [212.72.41.
89]
11 226 ms 207 ms 208 ms 4.68.63.42
12 224 ms 224 ms 224 ms 195.219.150.42
13 207 ms 206 ms 207 ms 82.197.196.146
14 208 ms 208 ms 208 ms www.tweaknews.eu [82.197.196.59]

Here is one on a different ISP network :
C:\Users\jesseg>tracert www.tweaknews.eu

Tracing route to www.tweaknews.eu [82.197.196.59]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 10.0.0.2
2 8 ms 7 ms 7 ms 196-210-128-129.dynamic.isadsl.co.za [196.210.12
8.129]
3 9 ms 13 ms 9 ms cdsl2-umh-gi0-0-0-3202.ip.isnet.net [196.38.74.2
33]
4 10 ms 10 ms 9 ms 196.26.210.204
5 20 ms 19 ms 19 ms 196.38.74.178
6 22 ms 23 ms 23 ms core1b-pkl-te0-0-0-0.ip.isnet.net [196.26.0.62]

7 210 ms 189 ms 187 ms 168.209.246.65
8 185 ms 185 ms 188 ms te0-7-0-1.ccr21.lon02.atlas.cogentco.com [149.6.
148.129]
9 187 ms 187 ms 189 ms be2328.ccr21.lon01.atlas.cogentco.com [130.117.4
9.85]
10 192 ms 192 ms 192 ms tinet.lon01.atlas.cogentco.com [130.117.15.22]
11 200 ms 201 ms 200 ms xe-1-0-7.ams12.ip4.gtt.net [89.149.180.145]
12 198 ms 195 ms 194 ms a2b-internet-gw.ip4.gtt.net [141.136.97.10]
13 201 ms 200 ms 200 ms fo1-asd-ve138.tc1.cambrium.net [82.197.196.106]

14 193 ms 193 ms 193 ms www.tweaknews.eu [82.197.196.59]

Trace complete.

C:\Users\jesseg>

 

[Ripstorm]

Hey AfriMan

Did you guys change anything with regards to twitch streaming? I was able to stream full 1080p this morning for the first time without any issues. I've already cancelled my Afrihost account at the beginning of the month, but will definitely renew it if the twitch streaming stays like this.

Thanks

 

[AfriMan]

Here is tracert on Afrihost network :
C:\Users\jesseg>tracert www.tweaknews.eu

Tracing route to www.tweaknews.eu [82.197.196.59]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 10.0.0.2
2 8 ms 7 ms 7 ms 105-236-0-65-esr-lo.mtnbusiness.co.za [105.236.0
.65]
3 17 ms 17 ms 28 ms 41.181.221.218
4 22 ms 16 ms 18 ms rb-dca-2.za--qux-b.za.mtnbusiness.net [41.181.16
5.118]
5 208 ms 208 ms 207 ms xp-dca-1.za--xp-cr-1.za-a.mtnns.net [196.44.0.15
5]
6 208 ms 208 ms 208 ms jh-cr-1.za--rb-cr-2.za.mtnns.net [196.44.0.42]
7 230 ms 215 ms 215 ms ls-cr-2.uk--rb-cr-1.za-g.mtnns.net [196.44.31.11
4]
8 217 ms 217 ms 230 ms am-cr-1.nl--ls-cr-2.uk-a.mtnns.net [196.44.31.18
3]
9 224 ms 223 ms 224 ms am-tpr-1.nl--am-cr-1.nl-a.mtn.net [209.212.111.1
41]
10 223 ms 223 ms 233 ms xe-4-1-0.edge5.Amsterdam1.Level3.net [212.72.41.
89]
11 226 ms 207 ms 208 ms 4.68.63.42
12 224 ms 224 ms 224 ms 195.219.150.42
13 207 ms 206 ms 207 ms 82.197.196.146
14 208 ms 208 ms 208 ms www.tweaknews.eu [82.197.196.59]

Here is one on a different ISP network :
C:\Users\jesseg>tracert www.tweaknews.eu

Tracing route to www.tweaknews.eu [82.197.196.59]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 10.0.0.2
2 8 ms 7 ms 7 ms 196-210-128-129.dynamic.isadsl.co.za [196.210.12
8.129]
3 9 ms 13 ms 9 ms cdsl2-umh-gi0-0-0-3202.ip.isnet.net [196.38.74.2
33]
4 10 ms 10 ms 9 ms 196.26.210.204
5 20 ms 19 ms 19 ms 196.38.74.178
6 22 ms 23 ms 23 ms core1b-pkl-te0-0-0-0.ip.isnet.net [196.26.0.62]

7 210 ms 189 ms 187 ms 168.209.246.65
8 185 ms 185 ms 188 ms te0-7-0-1.ccr21.lon02.atlas.cogentco.com [149.6.
148.129]
9 187 ms 187 ms 189 ms be2328.ccr21.lon01.atlas.cogentco.com [130.117.4
9.85]
10 192 ms 192 ms 192 ms tinet.lon01.atlas.cogentco.com [130.117.15.22]
11 200 ms 201 ms 200 ms xe-1-0-7.ams12.ip4.gtt.net [89.149.180.145]
12 198 ms 195 ms 194 ms a2b-internet-gw.ip4.gtt.net [141.136.97.10]
13 201 ms 200 ms 200 ms fo1-asd-ve138.tc1.cambrium.net [82.197.196.106]

14 193 ms 193 ms 193 ms www.tweaknews.eu [82.197.196.59]

Trace complete.

C:\Users\jesseg>

Every ISP will employ their own routing. I don't see anything wrong with our trace. If there was a spike in local latency, your end latency would be doubled, but your end latency is pretty consistent.

 

[JesseG]

Every ISP will employ their own routing. I don't see anything wrong with our trace. If there was a spike in local latency, your end latency would be doubled, but your end latency is pretty consistent.

OK, so how come when I start my online gaming, the NAT type is strict with AH but with a different ISP, the NAT type changes to moderate... Is this still a routing issue? Like I said before I am not that all that technically minded but there is difference between the 2 networks, one takes about 2 seconds to connect to a game while AH takes about 10 seconds to connect to the same game...

 

[AfriMan]

OK, so how come when I start my online gaming, the NAT type is strict with AH but with a different ISP, the NAT type changes to moderate... Is this still a routing issue? Like I said before I am not that all that technically minded but there is difference between the 2 networks, one takes about 2 seconds to connect to a game while AH takes about 10 seconds to connect to the same game...

NAT is network address translation, which doesn't really give much indication regarding performance. That's usually a setting that is either on or off - if we're talking about the same thing. Looking at the traces you posted, the end latency is different by 8ms, which is pretty marginal

 

[AfriMan]

Hey AfriMan

Did you guys change anything with regards to twitch streaming? I was able to stream full 1080p this morning for the first time without any issues. I've already cancelled my Afrihost account at the beginning of the month, but will definitely renew it if the twitch streaming stays like this.

Thanks

We have implemented changes on our systems which could already be taking effect. Glad that you're getting great service. Would be great if you post your stats for local and international throughput.

 

[Cynicster]

Uhhh... yeah that looks pretty normal to me. Not sure what your complaint is? It says SA, but it is cleary in Europe, since the 2nd last to last ping/hop is only 5ms. Logic dictates that NO data will travel from SA to Europe in 5 milliseconds.

Apologies for misinterpreting where the jump to EU takes place in my trace route; I see my mistake.
However, the MTN network does have much higher latency during the jump to EU, easily 20-40ms compared to other ISPs. 20-40ms might not be much when browsing/streaming but it definitely makes a world of difference when gaming.

I'd also like to point out that we're getting more professional feedback/diagnostics from people on this forum (for free) than the critical care staff of Afrihost who only seem to be able to provide pre-recorded responses to anything with say to them once given facts. In other words, these are people who are getting paid to provide an escalated support service, yet default to first line support type responses such as "Have you restarted your router?", "Have you tried a capped product?", "Have you requested a port reset from Telkom?" when confronted with irrefutable proof that the issue is on their network (I'm not referring to my 200ms hop issue here, that was my bad).