Afrihost overseas high pings?

BlackStatic said:
You should add a disclaimer that this is a high latency service.

Your Vuma based service has dramatically less ping, think 180ms vs 220ms, it's a shame because the Vuma uptime is horrible.
Click to expand...

Hi

The latency is still within acceptable ranges.
 
BlackStatic said:
Oh definitely, if you only watch youtube videos and check email. 220ms to Germany over fiber is embarrassing.
Click to expand...
I recommended you move to Openserve because I would never have imagined your ping would be 20ms more than mine seeing as you're only the next town over.

Your ping to JHB is more than my ping to CPT, so your options really are try and see if Openserve will do anything about it which I doubt, move back to Vumatel and the lower latency but sit with fibre that breaks every other week or live with the 20ms higher and have internet that works.
 
Soul Assassin said:
I recommended you move to Openserve because I would never have imagined your ping would be 20ms more than mine seeing as you're only the next town over.

Your ping to JHB is more than my ping to CPT, so your options really are try and see if Openserve will do anything about it which I doubt, move back to Vumatel and the lower latency but sit with fibre that breaks every other week or live with the 20ms higher and have internet that works.
Click to expand...
The sad thing is those figures are my Openserve ones, the Vuma ones were very low, true to fiber, but their actual availability/uptime in the Vaal region is horrible, you get constant service outages, had to move away.

I'll take the high latency with the high availability, though, it just doesn't seem to be necessary to be this high, especially local pings, you know.
 
BlackStatic said:
The sad thing is those figures are my Openserve ones, the Vuma ones were very low, true to fiber, but their actual availability/uptime in the Vaal region is horrible, you get constant service outages, had to move away.

I'll take the high latency with the high availability, though, it just doesn't seem to be necessary to be this high, especially local pings, you know.
Click to expand...
When I moved from Vuma to Openserve my ping went from 4 to 9ms so I said that's fair enough but yours went from 6-8ms? to 26ms, that's rough.

But yeah like you say Vumatel here is horrible I couldn't do it anymore, it was offline constantly.
 
Soul Assassin said:
When I moved from Vuma to Openserve my ping went from 4 to 9ms so I said that's fair enough but yours went from 6-8ms? to 26ms, that's rough.

But yeah like you say Vumatel here is horrible I couldn't do it anymore, it was offline constantly.
Click to expand...
Very sad, as performance wise their infrastructure was bloody impressive, never seen pings that low in my life, but yeah, no use having a Ferrari with no petrol.

Big thing is we use internet different now, our TVs run off internet etc, DSTV, live sports, so you can't just have random outages like the old ADSL days, only thing that sucked back then is you got disconnected from your online game lol, but this new stuff must be pumping 24/7. The whole requirement around constant availability is becoming a bigger issue now I think.

There is also loadshedding, not the loadhsedding itself, but the psychological knock-on effects, I noticed that people, myself included, have a VERY short fuse now with service reliability. It doesn't take much to make me outright leave a service, any service, that shows signs of interruption or degradation. We are all in survival mode now, things must be more dependable than Eskom, you know?
 
Last edited:
BlackStatic said:
I'm trying to play some BF3 which has a server in Germany but how is it that the EU servers there are 200-220ms ping on Openserve Afrihost? Surely it shouldn't be this bad?

Look at this crap

Tracing route to 94.250.199.214 over a maximum of 30 hops

1 3 ms 3 ms 2 ms homerouter.cpe [192.168.8.1]
2 16 ms 16 ms 17 ms 41.170.75.1
3 37 ms 27 ms 28 ms 165-73-14-41.ip.afrihost.co.za [165.73.14.41]
4 23 ms 23 ms 23 ms 169-1-21-28.ip.afrihost.co.za [169.1.21.28]
5 24 ms 23 ms 24 ms 169-1-21-232.ip.afrihost.co.za [169.1.21.232]
6 25 ms 23 ms 23 ms 169-1-21-233.ip.afrihost.co.za [169.1.21.233]
7 24 ms 24 ms 24 ms 206.249.2.73
8 211 ms 210 ms 211 ms be2385.ccr21.lon01.atlas.cogentco.com [154.54.40.93]
9 219 ms 242 ms 218 ms ldn-b3-link.ip.twelve99.net [62.115.9.28]
10 210 ms 211 ms 210 ms ldn-bb1-link.ip.twelve99.net [62.115.120.74]
11 218 ms 218 ms 218 ms prs-bb1-link.ip.twelve99.net [62.115.135.25]
12 232 ms 233 ms 233 ms ffm-bb1-link.ip.twelve99.net [62.115.123.12]
13 233 ms 233 ms 234 ms ffm-b5-link.ip.twelve99.net [62.115.114.89]
14 221 ms 222 ms 222 ms ae10-0.fra10.core-backbone.com [62.115.9.226]
15 221 ms 221 ms 278 ms cbb.fra14.edge-1.et-0-0-1 [5.56.21.154]
16 221 ms 221 ms 220 ms core01.fra.ix.fra14.ae2.0 [94.250.214.1]
17 222 ms 222 ms 222 ms
Click to expand...
WebAfrica on MetroFibre.

1 <1 ms <1 ms <1 ms 192.168.1.1
2 * * * Request timed out.
3 3 ms 3 ms 5 ms ethernet8-100.terdbn-igw1.net.echosp.link [102.67.179.3]
4 2 ms 3 ms 4 ms 100.127.5.2
5 1 ms 2 ms 3 ms 102.67.179.160
6 4 ms 4 ms 4 ms ntrce-os-cer-1-wan.osnet.co.za [196.25.130.69]
7 157 ms 157 ms 157 ms 10.189.30.6
8 187 ms 175 ms 176 ms 213.242.116.161
9 * * * Request timed out.
10 183 ms 182 ms 182 ms ae0-3356.lon10.core-backbone.com [212.113.8.42]
11 191 ms 191 ms 191 ms ae16-2074.fra10.core-backbone.com [81.95.2.137]
12 187 ms 187 ms 188 ms cbb.fra14.edge-1.et-0-0-1 [5.56.21.154]
13 * 190 ms 188 ms core01.fra.ix.fra14.ae2.0 [94.250.214.1]
14 193 ms 190 ms 192 ms spine1.fra.iw.xe-0-0-30.0 [94.250.214.251]
15 190 ms 191 ms 187 ms 94.250.199.214
 
Vorastra said:
WebAfrica on MetroFibre.

1 <1 ms <1 ms <1 ms 192.168.1.1
2 * * * Request timed out.
3 3 ms 3 ms 5 ms ethernet8-100.terdbn-igw1.net.echosp.link [102.67.179.3]
4 2 ms 3 ms 4 ms 100.127.5.2
5 1 ms 2 ms 3 ms 102.67.179.160
6 4 ms 4 ms 4 ms ntrce-os-cer-1-wan.osnet.co.za [196.25.130.69]
7 157 ms 157 ms 157 ms 10.189.30.6
8 187 ms 175 ms 176 ms 213.242.116.161
9 * * * Request timed out.
10 183 ms 182 ms 182 ms ae0-3356.lon10.core-backbone.com [212.113.8.42]
11 191 ms 191 ms 191 ms ae16-2074.fra10.core-backbone.com [81.95.2.137]
12 187 ms 187 ms 188 ms cbb.fra14.edge-1.et-0-0-1 [5.56.21.154]
13 * 190 ms 188 ms core01.fra.ix.fra14.ae2.0 [94.250.214.1]
14 193 ms 190 ms 192 ms spine1.fra.iw.xe-0-0-30.0 [94.250.214.251]
15 190 ms 191 ms 187 ms 94.250.199.214
Click to expand...
That's not bad but that one oke who is probably from Cape Town on Cool Ideas is hugging that server with his face with that 150-160ms lol, amazing what fiber can achieve.

I wonder what their secret sauce is.
 
BlackStatic said:
That's not bad but that one oke who is probably from Cape Town on Cool Ideas is hugging that server with his face with that 150-160ms lol, amazing what fiber can achieve.

I wonder what their secret sauce is.
Click to expand...
I want to know where that ~157ms hop is, on my trace, and why it jumps 20ms after that.
 
BlackStatic said:
That's not bad but that one oke who is probably from Cape Town on Cool Ideas is hugging that server with his face with that 150-160ms lol, amazing what fiber can achieve.

I wonder what their secret sauce is.
Click to expand...

Im from PE, 10 ms away from CPT, if I was in CPT my ping would have been even lower (around 155ms)

Also WebSquad also has good latency, 153ms from their CPT core:
Code: 
 1 160.119.238.113                    0%    2   0.2ms     0.3     0.2     0.3
 2 100.99.197.61                      0%    2   0.4ms     0.5     0.4     0.6
 3 160.119.232.19                     0%    2   0.5ms     0.6     0.5     0.7
 4 197.148.68.249                     0%    2   0.5ms     0.7     0.5     0.8
 5 197.148.71.54                      0%    2 140.9ms   140.9   140.9   140.9
 6 212.187.165.209                    0%    1 141.3ms   141.3   141.3   141.3
 7                                  100%    1 timeout
 8 212.113.8.42                       0%    1 163.1ms   163.1   163.1   163.1
 9 81.95.2.137                        0%    1 171.3ms   171.3   171.3   171.3
10 5.56.21.154                        0%    1 153.5ms   153.5   153.5   153.5
11                                  100%    1 timeout
12 94.250.214.239                     0%    1 154.1ms   154.1   154.1   154.1
13 94.250.199.214                     0%    1 153.4ms   153.4   153.4   153.4

From their JHB Core: 174ms.

They have a looking glass where you can run traces from their network: https://lg.as328137.net/
 
Jason-ZA said:
Im from PE, 10 ms away from CPT, if I was in CPT my ping would have been even lower (around 155ms)

Also WebSquad also has good latency, 153ms from their CPT core:
Code: 
 1 160.119.238.113                    0%    2   0.2ms     0.3     0.2     0.3
 2 100.99.197.61                      0%    2   0.4ms     0.5     0.4     0.6
 3 160.119.232.19                     0%    2   0.5ms     0.6     0.5     0.7
 4 197.148.68.249                     0%    2   0.5ms     0.7     0.5     0.8
 5 197.148.71.54                      0%    2 140.9ms   140.9   140.9   140.9
 6 212.187.165.209                    0%    1 141.3ms   141.3   141.3   141.3
 7                                  100%    1 timeout
 8 212.113.8.42                       0%    1 163.1ms   163.1   163.1   163.1
 9 81.95.2.137                        0%    1 171.3ms   171.3   171.3   171.3
10 5.56.21.154                        0%    1 153.5ms   153.5   153.5   153.5
11                                  100%    1 timeout
12 94.250.214.239                     0%    1 154.1ms   154.1   154.1   154.1
13 94.250.199.214                     0%    1 153.4ms   153.4   153.4   153.4

From their JHB Core: 174ms.

They have a looking glass where you can run traces from their network: https://lg.as328137.net/
Click to expand...
Outstanding.

I wonder why Liquid Telecom is so much worse?
 
DYreX146 said:
Looks like they're using Openserve transit which would go via Seacom which means the 157ms hop would be Marseille.
Click to expand...

Openserve doesn't use Seacom so it would be EASSy and WACS and the SAFE section for Asia.

That said the forward path is Openserve but the reverse could be Liquid Telecoms. Just like with our network. In CPT and DUR we have the forward path set as Openserve and in JHB forward path is Cogent but the reverse can be either Liquid or Openserve for KZN and CPT and Liquid, Openserve or Coghent on the reverse for JHB.

You might even find that you go out over WACS via London but come back via France to JHB.

It depends on what the European networks want to use to send you traffic. Even if you prepend the AS path those networks simply ignore it and it makes sense. Why do they need to send the traffic to you via LINX which will cost them transit when they can see you in AMX-IX or DE-CIX which is next to their network?

This is the biggest issue we are facing. So the only way to control it is to control the way European peers see our AS and that becomes expensive and basically unfeasible.
 
I've also been seeing higher pings than usual. Used to get around 173ms to League of Legends EUW from Pretoria on Afrihost. Now I'm getting 204ms on average this week. That extra 30ms goes a long way.

I'll check Wireshark tonight to see exactly which servers it's connecting to, but using the generic 104.160.141.3 "EUW" IP I see the following right now:

Afrihost:
1678364543803.png

CloudFlare WARP+:
1678364581891.png

Unfortunately, WARP seems to get more packet loss or packets out of order - the game just feels like something is wrong.

I posted about how it used to alternate higher and lower pings game-to-game before:

Higher ping to some international servers because of routing?

When I play League of Legends (laugh it up), every second game I get ±20ms higher ping than the other. One game will be 190ms, the next will be 205-210ms. This isn't just occasional, it happens maybe 90% off the time, and it lasts the entire game. When I run with Cloudflare WARP on, it...
mybroadband.co.za

But now it's just every game at 200ms+.
 
AfriNatic said:
Openserve doesn't use Seacom so it would be EASSy and WACS and the SAFE section for Asia.

That said the forward path is Openserve but the reverse could be Liquid Telecoms. Just like with our network. In CPT and DUR we have the forward path set as Openserve and in JHB forward path is Cogent but the reverse can be either Liquid or Openserve for KZN and CPT and Liquid, Openserve or Coghent on the reverse for JHB.

You might even find that you go out over WACS via London but come back via France to JHB.

It depends on what the European networks want to use to send you traffic. Even if you prepend the AS path those networks simply ignore it and it makes sense. Why do they need to send the traffic to you via LINX which will cost them transit when they can see you in AMX-IX or DE-CIX which is next to their network?

This is the biggest issue we are facing. So the only way to control it is to control the way European peers see our AS and that becomes expensive and basically unfeasible.
Click to expand...
Oops I guess I assumed too much, thanks for the info. I guess this is why CISP uses the express route via WACS and then peers with Cogent in London.
 
Vorastra said:
WebAfrica on MetroFibre.

1 <1 ms <1 ms <1 ms 192.168.1.1
2 * * * Request timed out.
3 3 ms 3 ms 5 ms ethernet8-100.terdbn-igw1.net.echosp.link [102.67.179.3]
4 2 ms 3 ms 4 ms 100.127.5.2
5 1 ms 2 ms 3 ms 102.67.179.160
6 4 ms 4 ms 4 ms ntrce-os-cer-1-wan.osnet.co.za [196.25.130.69]
7 157 ms 157 ms 157 ms 10.189.30.6
8 187 ms 175 ms 176 ms 213.242.116.161
9 * * * Request timed out.
10 183 ms 182 ms 182 ms ae0-3356.lon10.core-backbone.com [212.113.8.42]
11 191 ms 191 ms 191 ms ae16-2074.fra10.core-backbone.com [81.95.2.137]
12 187 ms 187 ms 188 ms cbb.fra14.edge-1.et-0-0-1 [5.56.21.154]
13 * 190 ms 188 ms core01.fra.ix.fra14.ae2.0 [94.250.214.1]
14 193 ms 190 ms 192 ms spine1.fra.iw.xe-0-0-30.0 [94.250.214.251]
15 190 ms 191 ms 187 ms 94.250.199.214
Click to expand...
Afrihost, Metro in PE
cmd_4sQgS5Pcr1.png
 
your first few hops seem to already be quite high

I'm on AH OS in CT

Code: 
michael@qnux:~$ traceroute -I 94.250.199.214
traceroute to 94.250.199.214 (94.250.199.214), 30 hops max, 60 byte packets
1 router.lan (192.168.88.1) 0.566 ms 0.569 ms 0.697 ms
2 41.170.73.33 (41.170.73.33) 2.036 ms 2.104 ms 2.112 ms
3 165-73-14-137.ip.afrihost.co.za (165.73.14.137) 5.586 ms 165-73-14-113.ip.afrihost.co.za (165.73.14.113) 5.692 ms 165-73-14-137.ip.afrihost.co.za (165.73.14.137) 5.595 ms
4 169-1-21-148.ip.afrihost.co.za (169.1.21.148) 3.334 ms 3.453 ms 3.429 ms
5 169-1-21-132.ip.afrihost.co.za (169.1.21.132) 5.104 ms 5.142 ms 5.239 ms
6 169-1-21-133.ip.afrihost.co.za (169.1.21.133) 3.005 ms 2.003 ms 2.739 ms
7 wdcnw-os-cer-1-wan.osnet.co.za (196.25.158.241) 13.612 ms 3.159 ms 3.074 ms
8 10.189.30.2 (10.189.30.2) 152.066 ms 151.930 ms 151.840 ms
9 ae72.edge7.London1.Level3.net (217.163.128.225) 151.769 ms 151.735 ms 151.795 ms
10 * * *
11 ae0-3356.lon10.core-backbone.com (212.113.8.42) 178.068 ms 178.025 ms 178.059 ms
12 ae16-2074.fra10.core-backbone.com (81.95.2.137) 182.939 ms 182.961 ms 182.898 ms
13 cbb.fra14.edge-1.et-0-0-1 (5.56.21.154) 182.410 ms 183.083 ms 183.132 ms
14 core01.fra.ix.fra14.ae2.0 (94.250.214.1) 183.437 ms 183.406 ms 183.348 ms
15 spine2.fra.iw.xe-0-0-30.0 (94.250.214.239) 183.607 ms 183.717 ms 183.6
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X