Please don't make the whole internet basically cloudflare. they've banned my VPN endpoint (Hetzner server) and as a result a huge chunk of websites already don't work for me despite me having done nothing wrong. I've heard reports of Tor users being restricted as well.

I don't foresee use of a CDN saving them much or any money. Cache hit ratio would probably be relatively low and the amount of storage required at the edge to cover an appreciable portion of traffic would be very expensive. E.g. their Akamai bill would be in the five figures per month if not six figures, depending on what kind of volume discount they could arrange, just based on outbound bandwidth. They're not the cheapest but also not the most expensive.

Serving up that large of a media library at mid scale just isn't really a great use case for a CDN, those that have to due to transit costs becoming truly enormous (e.g. Netflix) make an enormous investment in hardware at the edge that probably isn't affordable to IA (or necessary at this point).

Sibling comments are exactly correct -- we're a library. We don't track or record the activity of our patrons (see e.g. https://archive.org/services/docs/api/views.html#footnote-wh... ), and since most CDNs cannot offer the same guarantee, we can't use them.

At the bandwidth levels they are using they would need to use Cloudflare Enterprise, and in my experience that is way more expensive than other CDN providers.

Also, since archive.org has so much content, the caching ratio is going to be very bad and kill CDN efficiency while still requiring lots of direct bandwidth. Cheap direct bandwidth in their case looks best(which is what they seem to be doing).

They serve everything in-house and are their own AS; this allows them to provide stronger guarantees of reader privacy than sites that use a CDN.

The geographic distribution benefits for archive.org seem minimal to me - for the actual content high latency isn't a big deal.

It could be philosophical