Re: [arch-d] Comments on draft-lazanski-consolidation

[Christian Huitema <[e-mail]>]

On 11/12/2020 7:00 PM, Watson Ladd wrote:

> ECH is I think probably the draft's strongest argument. Any sort of
> anonymity service benefits from having more usage, so them that has,
> gets. This is why e.g. Tor is prefered over less popular solutions.
> It's worth considering these incentives but I think that there is, and
> always has been, a variety of shared hosting options. There are some
> very real technical difficulties here, and I think all of us would
> welcome well thought out proposals. But doing nothing about SNI
> leakage is not an option.


Watson's point is confirmed by market analysis. There is significant
competition in the CDN market -- the ascension of Fastly from 'small
startup" to "significant player" shows that. In fact, there is also
competition in the "cloud hosting" market -- AWS and Azure are big, but
there is a variety of other players, from Google Cloud to IBM, Oracle,
Linode, Digital Ocean and many others. I did an analysis of that here:
https://huitema.wordpress.com/2020/08/09/can-internet-services-hide-in-crowds/,
but there is better data available here:
https://www3.cs.stonybrook.edu/~mikepo/papers/esni.asiaccs20.pdf.

As for ECH, the analysis shows that most of these "top million" sites
are running on dedicated IP addresses, even when serviced by VPNs. Yes,
ECH requires "shared address" hosting to work, but no, there is no
evidence that big services are leveraging shared address provisioning to
grab market share.

-- Christian Huitema