Internet > Interesting Websites

Record for exaggeration?

(1/1)

Weaver:
The speedsmart speed tester came up with an upstream speed result of 2.12 Mbps today for me. (speed of what ? TCP payload?)

* 1904000 bps = The sum of the upstream sync rates
* 1683961 bps = Assuming 88.44% less protocol overheads of PPP and all lower layers IP PDU size 1500; 32 bytes of one-off, length-independent, per-PDU overhead ADSL2 ATM PPPoEoA
* 1616603 bps - the above at my chosen modem loading factor of 96%
So what’s that ? - 30% into the impossible?

burakkucat:
It's at times like this, that looking at all the figures can send one completely   :crazy:

Chrysalis:
its ok for me, alot of these weird results are related to software issues, sometimes caused by a/v software intercepting the traffic, other times might be down to browser been used, or lack of cpu grunt messing up packet timings,even things like line bonding which causes packet ordering issues may even have an impact.

For me one speedtester that started giving issues on the upstream was the dslreports test, but in the advanced settings page you can change the type of method used to transfer the data and that fixed it, which proves software implementations can cause bad results.

For a proper simple test maybe just run iperf instead? ;)

Weaver:
The reason has to be that they they truly hate IP bonding and I must be producing a lot of packets that are out of sequence.

You’ve given me an idea - I could run iperf on my external hosted Raspberry Pi ?

burakkucat:

--- Quote from: Weaver on April 13, 2019, 02:50:41 AM ---. . . I could run iperf on my external hosted Raspberry Pi ?

--- End quote ---

In theory, yes, you could.

Navigation

[0] Message Index