Capacity of iBurst's radio & distribution network loaded?

[Roman4604]

Was just doing some tests from the office to my router at home.

It seems the difference in latency between between their LNS router (the one that terminates your PPP session) and my router (other end of PPP) is huge.

This possibly shows that their radio & distribution (towers to core) nets must be pretty loaded, as my signal is always 100% (< -70 dB sig strength). Ping comparisons ...

To the LNS router

Pinging 196.30.31.100 with 32 bytes of data:

Reply from 196.30.31.100: bytes=32 time=9ms TTL=244
Reply from 196.30.31.100: bytes=32 time=11ms TTL=244
Reply from 196.30.31.100: bytes=32 time=30ms TTL=244
Reply from 196.30.31.100: bytes=32 time=10ms TTL=244
Reply from 196.30.31.100: bytes=32 time=12ms TTL=244
Reply from 196.30.31.100: bytes=32 time=22ms TTL=244
Reply from 196.30.31.100: bytes=32 time=12ms TTL=244
Reply from 196.30.31.100: bytes=32 time=11ms TTL=244
Reply from 196.30.31.100: bytes=32 time=24ms TTL=244
Reply from 196.30.31.100: bytes=32 time=13ms TTL=244
Reply from 196.30.31.100: bytes=32 time=16ms TTL=244
Reply from 196.30.31.100: bytes=32 time=12ms TTL=244
Reply from 196.30.31.100: bytes=32 time=10ms TTL=244
Reply from 196.30.31.100: bytes=32 time=15ms TTL=244
Reply from 196.30.31.100: bytes=32 time=12ms TTL=244
Reply from 196.30.31.100: bytes=32 time=16ms TTL=244
Reply from 196.30.31.100: bytes=32 time=17ms TTL=244
Reply from 196.30.31.100: bytes=32 time=13ms TTL=244
Reply from 196.30.31.100: bytes=32 time=17ms TTL=244
Reply from 196.30.31.100: bytes=32 time=13ms TTL=244

Ping statistics for 196.30.31.100:
Packets: Sent = 20, Received = 20, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 9ms, Maximum = 30ms, Average = 14ms

To my router

Pinging 196.2.102.190 with 32 bytes of data:

Reply from 196.2.102.190: bytes=32 time=490ms TTL=115
Reply from 196.2.102.190: bytes=32 time=181ms TTL=115
Reply from 196.2.102.190: bytes=32 time=176ms TTL=115
Reply from 196.2.102.190: bytes=32 time=145ms TTL=115
Reply from 196.2.102.190: bytes=32 time=117ms TTL=115
Reply from 196.2.102.190: bytes=32 time=91ms TTL=115
Reply from 196.2.102.190: bytes=32 time=86ms TTL=115
Reply from 196.2.102.190: bytes=32 time=304ms TTL=115
Reply from 196.2.102.190: bytes=32 time=212ms TTL=115
Reply from 196.2.102.190: bytes=32 time=107ms TTL=115
Reply from 196.2.102.190: bytes=32 time=110ms TTL=115
Reply from 196.2.102.190: bytes=32 time=117ms TTL=115
Reply from 196.2.102.190: bytes=32 time=96ms TTL=115
Reply from 196.2.102.190: bytes=32 time=125ms TTL=115
Reply from 196.2.102.190: bytes=32 time=95ms TTL=115
Reply from 196.2.102.190: bytes=32 time=84ms TTL=115
Reply from 196.2.102.190: bytes=32 time=456ms TTL=115
Reply from 196.2.102.190: bytes=32 time=257ms TTL=115
Reply from 196.2.102.190: bytes=32 time=93ms TTL=115
Reply from 196.2.102.190: bytes=32 time=122ms TTL=115

Ping statistics for 196.2.102.190:
Packets: Sent = 20, Received = 20, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 84ms, Maximum = 490ms, Average = 173ms

Look at the difference in latency and jitter (variation in ping times).

 

[Crash]

Has anyone considered the fact that they may have put ICMP very low on bandwidth prioritisation. So you get crap pings, it might not actually mean anything.

The LNS routers are probably connected by copper through an ATM back bone. So if you are on Diginet or ADSL the first hop will be fast.

If you did the test during the day the network will be loaded anyway.

 

[Roman4604]

Has anyone considered the fact that they may have put ICMP very low on bandwidth prioritisation. So you get crap pings, it might not actually mean anything.

My general performance/quality mirrors my pings, so I havent observed that they put ICMP on a lower priority. A quick call to the Skype Ou Ma (echo123) usually confirms this (uses UDP).

The LNS routers are probably connected by copper through an ATM back bone. So if you are on Diginet or ADSL the first hop will be fast.

My point is that I've had a perception that the performance problems with WBS network are usually to do with upstream bw or lack thereof. This illustrates that the problems are more to do with their distribution net ... i.e. they're skimping on bw between the towers & their core net.

 

[P0tenc]

I get my 1mbps 80% of the time.....Pings are usually good except mon-wednesday that went past, but that's fixed now. I'm +/- 10km away from the tower and get 100% signal strenght. The one techie told me iburst derives its signal from reflected signals as well. If those signals contain errors it will send your latency up.MY latency is usually between 50ms and 100ms but spikes to 300ms every now and then(to www.iburst.co.za). Considering my distance from tower, it's good. How far are you from the tower Roman4604? If you use USB check your frame error rate.

Also have you considered the problem of your tower maybe being heavily loaded? There are many ppl who use iburst that will say they get their 1mbps with good pings. So if your theory is right then every one should be getting high pings and lower speed.

 

[Roman4604]

How far are you from the tower Roman4604?

I would guess on about 5Km

If you use USB check your frame error rate.

I do check every now and then and FER% does spike up to 10-15% ... not so good. I am using ext. directional antenna so not sure about relfections, but I've always felt theres maybe interference in my area.

Also have you considered the problem of your tower maybe being heavily loaded?

I thinks so ... the Kyalami tower was much better 3 months ago when I got it. Also my ping times settle to a nice steady 60-80ms at 2am so I think it is largely congestion related.