Re: [arch-d] Centralization or diversity

[Stephane Bortzmeyer <[e-mail]>]

On Wed, Nov 06, 2019 at 09:58:08AM +1100,
 Martin Thomson <[e-mail]> wrote 
 a message of 25 lines which said:

> But my experience with centralized services is that they aren't
> centralized in the fault tolerance sense.  If I look at the big
> services, that scale is only achieved with careful distributed
> systems design.

I disagree. These systems are generally very homogeneous (software
and, of course, human management) so they still have SPOF. An example
is
<https://blog.cloudflare.com/details-of-the-cloudflare-outage-on-july-2-2019/>.

Centralized management is good for the administrators (less work and
better results) but may be dangerous for robustness. (That's why I
don't like the whole idea of SDN.)

> This draft really wants to say something less negative: that it values diversity on multiple axes.  The draft covers:
> 
> * fault domains geopolitical organizational
> 
> To which I might add:
> 
> * geographic
> * implementation
> * infrastructure (power, network, hardware)

And software! Many DNS zones, for instance, can be completely taken
down, if there is a 0-day vulnerability allowing to DoS all servers,
which are of the same make and version.

> Finally, I don't like the emphasis on DNS in this document.  It only
> serves to sensationalize.

Well, I disagree with the analysis of the draft (centralization to a
few public resolvers have nothing to do with DoH, it existed before)
but the problem is real. Unfortunately, we don't lack examples of
centralization.