DISCLAIMER: Project in progress.
Google, Ookla, speed test providers are known to be prioritized by ISPs.
Show you fun numbers! See its says 600Mbps up! Cool! Most people will see “Okay I see i’m getting 600Mbps up!” and then feel good about their service.
But the real upload rate is different. Obviously its in their business interest to throttle and such, there’s that FCC Net Neutrality Rules (2017) Ajit Pai debacle.
Anyways the focus of this post is basically, under the current scheme we’re in, trying to calc the *actual* download/uplink as a “layman.”
At least as of Ventura 13.5, built in terminal command:
networkQuality -v==== SUMMARY ====
Uplink capacity: 87.910 Mbps (Accuracy: High)
Downlink capacity: 504.511 Mbps (Accuracy: High)
Responsiveness: Medium (230 RPM) (Accuracy: High)
Idle Latency: 30.917 milliseconds (Accuracy: High)
Interface: en0
Uplink bytes transferred: 164.638 MB
Downlink bytes transferred: 1.274 GB
Uplink Flow count: 20
Downlink Flow count: 12
Start: 8/4/23, 3:02:58 PM
End: 8/4/23, 3:03:19 PM
OS Version: Version 13.5 (Build 22G74)
You’ll notice a difference between public test sites, which ISPs are known to prioritize traffic to to fake their upload capacity for average consumers googling “speed test”.
In my google/ookla, I get 600 Mbps up.
Someone had a trick to ping speedtest.net to trick the ISP to think you’re doing a fast speed test. Even pinging speedtest.net + running a local check does seem to change results.
Router speed test
The router has up/down accuracy. This in incredible.
But, we’re back to the “ISP prioritization.” — it’s likely they have prioritized that.
So how do we really find out?
Why don’t we just upload a couple gigs and see how long that takes?
Just uploaded to dropbox a 2GB file ~ took about 20 seconds.
2,000MB / 20s = 100MB/s
Convert to megabits: 800Mb
So looks like uploads on dropbox are working at intended speeds.
Okay that napkin math checks out.
So despite the local network test showing 60Mbps, I can napkin math uploading to dropbox was at 800Mbps.
Summary?
Clearly ISPs want to prioritize tests, but it’s not necessarily nefarious. Clearly various local tooling can show low uplink due to ISP not prioritizing it. But it makes sense, they need to distribute bandwidth in a way that’s most efficient..Poking holes and pointing fingers isn’t really productive.
I’m glad I did a manual upload i.e. “a real 2GB load” was prioritized / run at ~800Mbps (on wifi mind ya.)
That tells me I’m getting the speeds when I need it, and I’ll just have to trust my internet overlords they have our best interests at heart. And to be honest, I was surprised that when I uploaded something I actually needed to vs a synthetic test, it DID perform 10x.
I feel good enough for now.
Yuji Tomita 