Web Squad ISP

Status
Not open for further replies.
Tag me if you share the results, I use rclone with gdrive more often than the official client

with rclone (and other clients) I find adjusting chunk sizes greatly improves performance as smaller chunks with the latencies we have here in RSA yield poor TCP-window/transfer curves. Think about the way the graph on Speedtest.net curves up as the test happens, same thing with any TCP transfer. So the longer you're running that chunk for, the better the speed.
 
Man, now I wanna cancel CISP and sign up for Websquad 1000/100. But, my VOIP is with CISP and worried about running into potential problems with my telephone lines if I move.
 
@ipv6_warrior

Looks like it was just Amazon taking a while to update routes. Check it out now

AWS return route UK & DE seems ok, although latency is still higher than what I'd think it should be?

Code:
# traceroute speedtest.cpt.websquad.co.za
traceroute to speedtest.cpt.websquad.co.za (160.119.233.123), 30 hops max, 60 byte packets
 1  ec2-52-56-0-132.eu-west-2.compute.amazonaws.com (52.56.0.132)  21.204 ms  21.177 ms ec2-52-56-0-134.eu-west-2.compute.amazonaws.com (52.56.0.134)  18.436 ms
 2  100.66.0.242 (100.66.0.242)  11.680 ms * 100.66.0.218 (100.66.0.218)  5.745 ms
 3  100.66.0.53 (100.66.0.53)  13.986 ms 100.66.0.127 (100.66.0.127)  11.110 ms 100.66.0.85 (100.66.0.85)  18.763 ms
 4  100.65.1.129 (100.65.1.129)  0.222 ms 100.65.1.33 (100.65.1.33)  0.229 ms 100.65.1.65 (100.65.1.65)  0.867 ms
 5  52.94.33.89 (52.94.33.89)  2.146 ms 52.94.33.81 (52.94.33.81)  2.070 ms 52.94.33.89 (52.94.33.89)  2.014 ms
 6  52.94.33.154 (52.94.33.154)  6.562 ms 52.94.33.152 (52.94.33.152)  2.440 ms 52.94.33.156 (52.94.33.156)  2.401 ms
 7  54.239.101.137 (54.239.101.137)  1.994 ms 54.239.101.159 (54.239.101.159)  2.141 ms  2.180 ms
 8  54.239.46.80 (54.239.46.80)  159.212 ms 54.239.46.82 (54.239.46.82)  158.133 ms 54.239.46.80 (54.239.46.80)  159.804 ms
 9  54.239.46.75 (54.239.46.75)  159.104 ms 54.239.46.73 (54.239.46.73)  158.099 ms  158.088 ms
10  54.239.46.74 (54.239.46.74)  158.436 ms 54.239.46.72 (54.239.46.72)  158.042 ms  158.431 ms
11  52.93.57.130 (52.93.57.130)  159.231 ms 52.93.57.114 (52.93.57.114)  158.848 ms 52.93.57.50 (52.93.57.50)  159.352 ms
12  52.93.57.11 (52.93.57.11)  158.192 ms 52.93.57.75 (52.93.57.75)  158.099 ms 52.93.57.57 (52.93.57.57)  159.905 ms
13  websquad.ixp.capetown (196.10.140.137)  157.882 ms  157.918 ms  157.829 ms
14  core.cr-01.cp1.za.ws.net.za (160.119.232.5)  157.992 ms  157.978 ms  157.953 ms
15  core.as-01.cp1.za.ws.net.za (160.119.233.190)  158.044 ms  158.024 ms  158.017 ms
16  speedtest.cpt.websquad.co.za (160.119.233.123)  158.044 ms  158.038 ms  158.030 ms

Code:
# ping speedtest.cpt.websquad.co.za
PING speedtest.cpt.websquad.co.za (160.119.233.123) 56(84) bytes of data.
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=1 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=2 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=3 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=4 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=5 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=6 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=7 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=8 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=9 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=10 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=11 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=12 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=13 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=14 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=15 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=16 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=17 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=18 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=19 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=20 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=21 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=22 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=23 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=24 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=25 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=26 ttl=49 time=158 ms
^C
--- speedtest.cpt.websquad.co.za ping statistics ---
27 packets transmitted, 26 received, 3% packet loss, time 26026ms
rtt min/avg/max/mdev = 158.042/158.123/158.356/0.172 ms
 
@websquadza what are your Vumatel contention ratios by the way?

If you're on GPON (Vumatel Aerial), you're going to get the inherent contention that comes from splitting a 1.25 Gbps port by x many homes. If you're on trenched, you have a dedicated 1 Gbps circuit to the nearest Vuma POP, each 24 port switch has 10G uplinks to the core. Then you have backhaul contention. Vumatel actively manages backhaul by making sure ports don't saturate and upgrades (generally) happen on time and before ports fill up. Then you have ISP contention. We don't work by the philosophy of buying X Gbps upstream bandwidth and sharing it between n many clients. We operate on a port basis across 4 datacentres with multiples of 10G ports into exchanges and upstream partners (and between facilities). When these ports reach a certain saturation level, we upgrade. So to actually nail down a contention ratio is almost impossible.

I hope this explanation makes sense?
 
AWS return route UK & DE seems ok, although latency is still higher than what I'd think it should be?

Code:
# traceroute speedtest.cpt.websquad.co.za
traceroute to speedtest.cpt.websquad.co.za (160.119.233.123), 30 hops max, 60 byte packets
1  ec2-52-56-0-132.eu-west-2.compute.amazonaws.com (52.56.0.132)  21.204 ms  21.177 ms ec2-52-56-0-134.eu-west-2.compute.amazonaws.com (52.56.0.134)  18.436 ms
2  100.66.0.242 (100.66.0.242)  11.680 ms * 100.66.0.218 (100.66.0.218)  5.745 ms
3  100.66.0.53 (100.66.0.53)  13.986 ms 100.66.0.127 (100.66.0.127)  11.110 ms 100.66.0.85 (100.66.0.85)  18.763 ms
4  100.65.1.129 (100.65.1.129)  0.222 ms 100.65.1.33 (100.65.1.33)  0.229 ms 100.65.1.65 (100.65.1.65)  0.867 ms
5  52.94.33.89 (52.94.33.89)  2.146 ms 52.94.33.81 (52.94.33.81)  2.070 ms 52.94.33.89 (52.94.33.89)  2.014 ms
6  52.94.33.154 (52.94.33.154)  6.562 ms 52.94.33.152 (52.94.33.152)  2.440 ms 52.94.33.156 (52.94.33.156)  2.401 ms
7  54.239.101.137 (54.239.101.137)  1.994 ms 54.239.101.159 (54.239.101.159)  2.141 ms  2.180 ms
8  54.239.46.80 (54.239.46.80)  159.212 ms 54.239.46.82 (54.239.46.82)  158.133 ms 54.239.46.80 (54.239.46.80)  159.804 ms
9  54.239.46.75 (54.239.46.75)  159.104 ms 54.239.46.73 (54.239.46.73)  158.099 ms  158.088 ms
10  54.239.46.74 (54.239.46.74)  158.436 ms 54.239.46.72 (54.239.46.72)  158.042 ms  158.431 ms
11  52.93.57.130 (52.93.57.130)  159.231 ms 52.93.57.114 (52.93.57.114)  158.848 ms 52.93.57.50 (52.93.57.50)  159.352 ms
12  52.93.57.11 (52.93.57.11)  158.192 ms 52.93.57.75 (52.93.57.75)  158.099 ms 52.93.57.57 (52.93.57.57)  159.905 ms
13  websquad.ixp.capetown (196.10.140.137)  157.882 ms  157.918 ms  157.829 ms
14  core.cr-01.cp1.za.ws.net.za (160.119.232.5)  157.992 ms  157.978 ms  157.953 ms
15  core.as-01.cp1.za.ws.net.za (160.119.233.190)  158.044 ms  158.024 ms  158.017 ms
16  speedtest.cpt.websquad.co.za (160.119.233.123)  158.044 ms  158.038 ms  158.030 ms

Code:
# ping speedtest.cpt.websquad.co.za
PING speedtest.cpt.websquad.co.za (160.119.233.123) 56(84) bytes of data.
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=1 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=2 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=3 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=4 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=5 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=6 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=7 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=8 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=9 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=10 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=11 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=12 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=13 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=14 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=15 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=16 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=17 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=18 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=19 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=20 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=21 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=22 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=23 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=24 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=25 ttl=49 time=158 ms
64 bytes from speedtest.cpt.websquad.co.za (160.119.233.123): icmp_seq=26 ttl=49 time=158 ms
^C
--- speedtest.cpt.websquad.co.za ping statistics ---
27 packets transmitted, 26 received, 3% packet loss, time 26026ms
rtt min/avg/max/mdev = 158.042/158.123/158.356/0.172 ms

I would have also thought we'd see about 149ms like we usually do to other services. But I can't speak for Amazon's network design here.

With regards to your suggestion about de-aggregating on transit - it lands up taking away from the benefit of various providers' superior peering. Telia for example has a far superior route to us via Cape town (fewer AS hops), whereas Zayo will see our JHB as a shorter, superior route. In addition, these are direct peers to our transit providers, so we can encourage transit providers to sort out capacity and packet loss issues that are on their network when they arise. We're bringing on a 3rd provider now that will definitely influence inbound routes significantly. Once that's in, we'll look at various inbound route strategies.
 
I would have also thought we'd see about 149ms like we usually do to other services. But I can't speak for Amazon's network design here.

With regards to your suggestion about de-aggregating on transit - it lands up taking away from the benefit of various providers' superior peering. Telia for example has a far superior route to us via Cape town (fewer AS hops), whereas Zayo will see our JHB as a shorter, superior route. In addition, these are direct peers to our transit providers, so we can encourage transit providers to sort out capacity and packet loss issues that are on their network when they arise. We're bringing on a 3rd provider now that will definitely influence inbound routes significantly. Once that's in, we'll look at various inbound route strategies.

Actually I've seen 141ms with CISP, outbound on Amazon, return via their LINX peering :)

As for the de-aggregating, IMHO if that saves 20ms on r/t it's better than any "superior peering" providers might have.
 
@websquadza
Got an edgerouter x, still only negotiating to 100mbps, ge2 negotiates to 1gbps.
Tried forcing to 1gbps, but it just drops the connection.
Logged a ticket ref #665833
 
Dug up @websquadza ASN https://bgp.he.net/AS328137
Will need to do some testing regarding routes (compared to something like Cool Ideas), but I know Work Online isn't much to write home about (WIRULink peers with them, and the pings are the bottom end of average), no idea about HE but I'm looking forward to finding out haha
 
Dug up @websquadza ASN https://bgp.he.net/AS328137
Will need to do some testing regarding routes (compared to something like Cool Ideas), but I know Work Online isn't much to write home about (WIRULink peers with them, and the pings are the bottom end of average), no idea about HE but I'm looking forward to finding out haha

So many South African ISPs use Seacom and HE as transit providers. Not sure why maybe HE is better than the rest?

Supersonic using the MTN business network obviously got a couple of different providers.
 
Man, now I wanna cancel CISP and sign up for Websquad 1000/100. But, my VOIP is with CISP and worried about running into potential problems with my telephone lines if I move.

Way not number transfer to and try WebSquad's VoIP??

I'll seriously advise people to rather consider a more "independent" provider like ConnectionTelecom for VoIP services
 
Status
Not open for further replies.
Top
Sign up to the MyBroadband newsletter
X