Tool: Dashboard Throughput test varies depending on device ?

[Thygesen]

Hi !

I seem to run into very different results using the "Dashboard Throughput" tool depending on the device type..

Setup:

ISP -> Switch1 -> Switch2

Switch1-> AP1

Switch1-> MX1

ISP: 1G up/downlink

So:

Switch1 (120-8FP) says: 20.1 Mbit

Switch2 (120-8LP) says: 19.9 Mbit

AP1 (MR33) says: 433.9 Mbit

MX1 (MX64) says: 308.4 Mbit

The same pattern is seen with other installations as well ..

How come, and if that's life, then what should I expect from these devices ?

Lotz of regards

Thyge

[ww]

My guess

Ap and mx have pretty good processor/chipset for software functions

Switch dashboard throughput always have  been low. Switches uses hardware/asic for processing packets and only need a slow processor, limiting the throughput tool.

But the Dashboard throughput tool is not designed for reliable testing wan speed

https://documentation.meraki.com/General_Administration/Tools_and_Troubleshooting/Throughput_test_to...

[Thygesen]

😀I see that, and why  ..

What I don't get is: What is the intention with the tool then ?
If it is a CPU dependent measure, it could be a % then .. ?

My expectation was not a WAN performance tool, but a management/Meraki cloud control meter ..

Regards
Thyge

[ww]

I think its  usefull as indication  for testing "low speed" (max 200 Mbit) wan links from mr / mx / mg if customers have issues with  wan speed.

I dont see any other use for it.

[Thygesen]

That makes some sort of reason 😀.
We were just confused since the numbers made no sense, when we looked on the topology ..

[MFagan]

I agree! If the feature won't give us useful troubleshooting info as built (because, perhaps of hardware limits) it should probably not be shown as an option on the other devices. In my example, I have a small campus with interconnected fiber links and was assuming my poor results from a switch/AP's which are downstream from the MX via Fiber might indicate bad fiber/transceivers but testing back at the site with teh MX and the cicuit revealed that ALL switches and access points under-report what they are actually getting from the MX. 

[Brash]

Interesting. My tests are quite similar.

ISP <-> MX <-> MS <-> AP

MX: 51mbps

MS: 10mbps

AP: 51mbps

This network has a 50mbps ISP uplink.

Assumedly the switches have some sort of rate limiter applied for dashboard communication?

Either way, Meraki make it very clear that the test is not a valid dataplane speed test and so I don't see a limited dashboard throughput being a problem.

[Thygesen]

🙂 Agree 🙂

I think I need to understand what the tool is meant for ..

I thought it was something like: What is the box experiencing of link expectation/performance from the ISP/Cloud.
What I see then is that it's more like "What can this CPU do with the link" ..
All the switches seems to have same degree of performance number ..

It just confused one of our remote tech's doing troubleshooting at some wireless installation ..

Client: Slow, AP: seems fine, SWITCH: slow .. ???

Regards

Thyge

[Thygesen]

🙂
Just for fun did a couple of tests on a private installation:

Setup: ISP->SW1->SW2->AP (daisy-chained)
Sw1: MS120-8FP: 18.2Mbps
Sw2: MS120-8LP: 20.7Mbps
AP: MR33: 428.4Mbps ..
🙂