MWeb peering link with Telkom/SAIX cut

[Franna]

I have just noticed that my pings from MWeb uncapped to Telkom/SAIX jumped from less than 20ms to more than 250ms - so MWeb is busy updating routing paths.

Pinging saix.net [196.25.1.200] with 32 bytes of data:
Reply from 196.25.1.200: bytes=32 time=251ms TTL=242
Reply from 196.25.1.200: bytes=32 time=249ms TTL=242
Reply from 196.25.1.200: bytes=32 time=249ms TTL=242
Reply from 196.25.1.200: bytes=32 time=249ms TTL=242
Ping statistics for 196.25.1.200:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 249ms, Maximum = 251ms, Average = 249ms

Pinging telkom.co.za [196.43.22.222] with 32 bytes of data:
Reply from 196.43.22.222: bytes=32 time=245ms TTL=243
Reply from 196.43.22.222: bytes=32 time=245ms TTL=243
Reply from 196.43.22.222: bytes=32 time=244ms TTL=243
Reply from 196.43.22.222: bytes=32 time=244ms TTL=243
Ping statistics for 196.43.22.222:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 244ms, Maximum = 245ms, Average = 244ms

 

[Terumo]

I am all for the free peering strategy, unfortunately we will suffer a bit in the short term. I noticed the increase in latency in MoH is lately 200+ where it was 24 and up :S We need more game servers Mweb

 

[luketang56]

All fine here.. Telkom to Mweb

Pinging mweb.co.za [196.2.63.110] with 32 bytes of data:
Reply from 196.2.63.110: bytes=32 time=54ms TTL=122
Reply from 196.2.63.110: bytes=32 time=52ms TTL=122
Reply from 196.2.63.110: bytes=32 time=52ms TTL=122
Reply from 196.2.63.110: bytes=32 time=55ms TTL=122

Ping statistics for 196.2.63.110:
Packets: Sent gra = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 52ms, Maximum = 55ms, Average = 53ms

 

[Gordon_R]

Pings from MWeb (CT) to SAIX are worse this afternoon (consistently 415ms). Looks like Telkom are reciprocating...

 

[Pada]

The story of Mweb: There and back again

Lekke. Mweb to Telkom traffic & vice versa are going via London, UK:

traceroute to www.telkom.co.za (196.43.22.222), 30 hops max, 60 byte packets
 1  172.30.0.129 (172.30.0.129)  1.611 ms  1.607 ms  1.603 ms
 2  172.30.30.1 (172.30.30.1)  1.808 ms  2.047 ms  2.096 ms
 3  41-134-xxx-xxx.dsl.mweb.co.za (41.134.xxx.xxx)  5.241 ms  5.518 ms  5.709 ms
 4  tengig-0-0-0.vic-lns-1.mweb.co.za (196.22.189.11)  41.435 ms  42.520 ms  44.506 ms
 5  vl-92.vic-hscore-2.mweb.co.za (196.22.189.3)  47.419 ms  50.721 ms  52.886 ms
 6  tengig-0-0-0-0-12.vic-up-1.mweb.co.za (196.22.169.241)  57.680 ms  45.439 ms  40.687 ms
 7  tengig-0-2-0-0.mid-1.mweb.co.za (196.22.169.219)  48.917 ms  49.231 ms  51.203 ms
 8  196.22.175.218 (196.22.175.218)  289.359 ms  292.031 ms  293.633 ms
 9  149.6.98.5 (149.6.98.5)  283.000 ms  288.896 ms 149.6.98.17 (149.6.98.17)  301.454 ms
10  te7-2.ccr01.lon01.atlas.cogentco.com (154.54.36.165)  300.395 ms  300.708 ms  301.018 ms
11  te2-2.mpd01.lon01.atlas.cogentco.com (130.117.1.74)  305.465 ms  305.832 ms  306.148 ms
12  te2-1.ccr01.lon05.atlas.cogentco.com (130.117.49.94)  278.436 ms  277.744 ms te1-1.ccr01.lon05.atlas.cogentco.com (130.117.49.106)  280.186 ms
13  149.6.2.194 (149.6.2.194)  283.114 ms  278.676 ms  279.441 ms
14  rrba-ip-esr-1-ge-6-0-0.telkom-ipnet.co.za (196.43.11.166)  459.845 ms  457.049 ms  458.629 ms
15  rrba-ip-bssr-1-ge-2-48.telkom-ipnet.co.za (196.43.23.6)  466.289 ms  470.955 ms  472.200 ms
16  nbsc-ip-bssr-2-atm-5-0-0-1.telkom-ipnet.co.za (196.43.23.34)  474.710 ms  476.343 ms  478.510 ms
17  nbsc-ip-rcache-1-vif2.telkom-ipnet.co.za (196.43.22.222)  494.327 ms  496.479 ms  500.195 ms

I hope people are raping SAIX's international b/w this way to such an extent that they will be forced to do local peering for free

 

[Gordon_R]

The story of Mweb: There and back again

Lekke. Mweb to Telkom traffic & vice versa are going via London, UK:

I hope people are raping SAIX's international b/w this way to such an extent that they will be forced to do local peering for free

Yeah, I did notice the double round trip! (Already posted a tracert on another thread, didn't want to cross post).

This does raise a lot of interesting questions (including that posed by Andrew Alston as to whether this counts as 'local-only' traffic), but it looks like the 'war' is escalating.

The broader question is which business model is more sustainable, since MWeb has 'unlimited' bandwidth (via SEACOM), and Telkom has to pay for every GByte via SAT-3...

 

[Pada]

This does raise a lot of interesting questions (including that posed by Andrew Alston as to whether this counts as 'local-only' traffic)

I'm also interested in the answer for that! My guess would be that you would actually be using international cap if you're connecting from SAIX to Mweb.
If that is the case, then RouteSentry users on SAIX will be screwd.