Storj, specifically using RClone and the native Uplink CLI (vs the S3 gateway). Super cheap, P2P, built-in client-side encryption are what keep me on it (and steering clear of the nightmare that is AWS).
Not really addressing the core issue but here’s some unsolicited advice: run some type of resolver locally that caches; specifically something that’ll cache all the NXDOMAIN and/or 0.0.0.0 you’ll get back. It’ll really speed things up especially if you can add in some prefetching. I do this with unbound.
So how many of these “experiments” do we need to have until we make some policies based on the results? I see this all the time, but it’s always just that: an “experiment”
If you’re at a Walmart a good trick if you’ve only got a handful of items is checking out in electronics, apparel or HBA/cosmetics.
But in huge cities those cashiers are oftentimes missing or bombarded with key duty (where they lock up items that cost like $2) so it’s hit or miss.
(Regarding locking up simple items like deodorant: if it takes 45 minutes for someone to come unlock it and the only way to get someone to come over with a key is by shouting “help” very loudly I’m ordering it on Amazon.)