We have a client implementing a new VOIP system and one of the tests is relating to bufferbloat which is coming back with a sub-par result.
Test Pkt Loss Avg Jitter Avg Latency
Idle 0/1495 (0.0%) 2 ms 11 ms
Downloading 1/1495 (0.1%) 4 ms 19 ms
Uploading 0/1495 (0.0%) 5 ms 237 ms
Avg bloat : 117.00 ms
TEST RESULT : Grade C (under 200 ms)
The above results are AFTER implementing QoS at the Client via GPO, the Switch, and the Watchguard Firewall, bypassing all Proxies, and having direct packet filters for the application.
- Geolocation Off
- IDS Off
- App Control Off
Internet connection in question is an NBN Fibre 100 Down, 40 Up, 3ms latency.
Existing VOIP services are running without an issue, so I'm not sure if this is a red herring and if it could be safely disregarded.
Has anyone come across this type of metric before? And any ideas on how to improve it?
Thanks in advance,
Uploading latency of 237 ms seems quite odd.
I would suggest seeing if you can point that test at any other server. If upload were that bad, your download would almost certainly also be bad under most circumstances.
WatchGuard Customer Support
Yeah I suspect that maybe their testing utility server is on a network much further away than their actual localized VOIP servers.
From looking at the software there is no way to change where it points to.
That's like the latency between Australia and the USA.. On a good day..
Adrian from Australia
There is a lot of ocean between Australia and the USA.