[Starlink] Widely different bufferbloat on two machines connected to one access point

Doc Searls doc at searls.com
Tue Jan 17 11:27:04 EST 2023


A question for the true geniuses here, before I call an Apple one who isn't...

We have two laptops here, both feet away from the nearest Eero hot spot in a household network fed by a cable modem, an Aris DOCSIS 3.1 router and a 24-port Netgear switch.

My laptop is a 6-year old MacBook Pro. My wife's laptop is a brand new MacBook Air that rocks in every respect but its network connection,

Using Fast.com, I'm getting 450 Mbps down, 38 Mbps up, with 14 ms of unloaded latency and 28 ms of loaded latency. My wife is getting 67 Mbps down, 28 Mbps up, 15 ms of unloaded latency and 780 ms of loaded latency. 

Repeated tests yield similar results.

In other words, her machine is getting a lot of bufferbloat and mine is not.

If I connect by Ethernet direct to the switch through a USB-C to Ethernet adapter, I get 880 Mbps down, with other values about the same. Using the same Ethernet connection, she gets roughly the same results as she gets on WiFi.

So it seems the problem is with her new machine. If so, what's the fix?

We do have an immediate need: downloading seven years of emails via IMAP from her server at Rackspace, which we are leaving because the company failed spectacularly in a ransomware attack and is unworthy of customer faith. It has taken most of a week so far to download 390,000 emails, and we would like to speed that up.

I'm less lucky, since my mail (unlike hers) was on the Rackspace Exchange server, which was the target of the attack. I have been unable to retrieve anything so far, and I am losing hope that I ever will.

Anyway, the question is about radically different bufferbloat on two machines connected the same way—and what to do about it.

Thanks,

Doc


More information about the Starlink mailing list