Yes, correct. In general in my calculations I took into account worst-case scenarios to see the worst-case result. Realistically I would believe this to be an opt-in, so way way way fewer websites have it, and not see a significant bandwidth increase from Tor to Cloudflare, because of the above. People who used TBB, will keep using TBB. No more people will probably come from this.

In general, the only thing I would worry about would be the HSDir size increase, NOT the bandwidth or circuits, or similar..

Also, I don’t think Cloudflare spent so much time in engineering, just to take down Tor.. If they really wanted to do this, they could have done it in “cheaper” ways.. They spent the time so this can work, so they already (probably) took into account the load that will be placed on the network, and determined it to be bearable.. (I hope so :P)

Antonis

On 2018-08-20 15:39, DaKnOb wrote:

HOWEVER, Cloudflare doesn’t need to hide their location. Everyone knows their servers. So they can use single hop Onion Services, and not the traditional three hop ones.

That means that in terms of total traffic, they will use 43 + (3*5) = ~ 60 Gb/s, out of the ~ 90 Gb/s available.

One thought that comes to mind: Doesn't this same traffic already likely flow through tor?

Right now browsing my personal blog from TBB uses 'x' number of bytes which pass through relays and an exit, won't the primary difference be that we no longer require an exit by routing directly to a Cloudflare hosted single hop onion service?

This attention might bring more users to using Tor which is a separate and valid concern in terms of anticipating growth.

I'm still waiting for Cloudflare to activate me on this beta program to be able to explore how well it actually works.
_______________________________________________
tor-relays mailing list
tor-relays@lists.torproject.org
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays