[LEDE-DEV] RFC: Throughput testing results.

[Ben Greear]

On 05/06/2016 10:20 AM, Toke Høiland-Jørgensen wrote:
> Ben Greear <greearb at candelatech.com> writes:
>
>> I am interested in feedback on the testing output. My goal is to add a
>> few more different hardware configurations and then do nightly (or at
>> least weekly) tests.
>
> So that is showing up to 10s of *seconds* of latency, right? (I'm not
> sure I'm reading the units right).

Yes, 10 seconds of latency.  My traffic generator is using pfifo-fast,
RENO, and default socket sizes, so it can be at least part of the problem.

>
> 4th graph: The title says "PDU/s" (what's that), but the labels on the
> graph says Pps.

protocol data units, so UDP traffic with 24k frames would be one pdu per
roughly 24k bytes, but there would be a lot more actual ethernet/wifi frames.

For the tests I plan to run, PDU/s is effectively the same as packets-per-second
for UDP, and it is somewhat irrelevant for TCP.  In general, if TCP is involved,
those numbers are a guess at best.

> What does the graph labelled "Stations requested UP vs Bringup time..."
> show?

X axis is number of stations for that iteration, and the bringup time is how
long it took to associate and get an IP address.  Long bringup times can mean
we are bringing up lots of stations, or DHCP is slow, or who knows what.  To really
understand any issue, it would take sniffing and/or looking at logs.

>
> Why is there such a distinct drop for TX link rate (in the last graph)
> for the second half of stations, but not for RX link rate?

Limitation in the firmware on the station emulator.  I actually should have
that fixed next time I update that system.  Also, not much upload traffic,
so rate-ctrl is not so important or accurate in upstream direction.

Thanks!
Ben


>
> -Toke
>


-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com