nummagumma2
asked on
Suggestions for accurate speedtest when link is using CEF
I have a circuit provided by a huge ISP that is 4 T1s bound using CEF (Cisco Express Forwarding). When using www.speedtest.net this circuit does not show the 6Mbps throughput that we would expect. Another vendor's circuit using MLPP for 4 bound T1s is showing 6Mbps throughput.
Is there an online test that will show the accurate results for the CEF circuit?
If you know this or can point to information, I'd be appreciate too:
Is there some problem with the way CEF works that would not provide a fully bonded circuit at the higher speed?
Is there an online test that will show the accurate results for the CEF circuit?
If you know this or can point to information, I'd be appreciate too:
Is there some problem with the way CEF works that would not provide a fully bonded circuit at the higher speed?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
For giggles, can you do two speed tests at the same time, and see if you then get >3Mbps down and >1.2Mbps up?
Getting 3Mbps may indicate that they have per-packet forwarding enabled on their end.
Getting 3Mbps may indicate that they have per-packet forwarding enabled on their end.
ASKER
I've been told they are using per-packet forwarding, but I don't know what that means in terms of the overall speed that I'd be getting. It sounds like you do... is there some reading I can do on this?
ASKER
Thanks guys. I learned from speedtest.net that they don't support speed tests on CEF - and that what you both said was true - CEF will show maximum throughput of less than the combined speed due to the way it deals with packets.
ASKER
The speedtests show something like 3Mbps down and 1.2 up, a far cry from the 6/6 I'd expect. It looks like a DSL connection, albeit a far more robust solution.
Thanks for the comments so far.