Re: [Sidrops] rfc8210bis further review - question 3

[Job Snijders <[e-mail]>]

Dear all,

Question 3
==========

Section 12 "Deployment Scenarios" - I'm not sure this section serves a
purpose in an already lengthy document.

For example, what's the difference between 'small' and 'large' multiomed
end sites? The section says "complex topologies are ill-advised" and
goes on to discuss caches fetching from each other rather than directly
from the RPKI, which is a complex topology because of lack of loop
detection?

Also a bit yucky how this section starts off with "may wish to outsource
the RPKI cache to one or more of their upstream ISPs", as this is not a
practise I recognize as widely used, or would recommend based on my
experience in the field. There are a handful of instances that I know
of, none of them seem optimal: a cloud provider offers a public RTR
service, they are not 'upstream', and the service is broken in subtle
ways that are very hard to troubleshoot. Some hobbyists offer RTR
servers sometimes in context of an IXP community, and this practise too
isn't 'upstream provider'.

I recommend removing section 12 in its entirety: it lacks sufficient
detail to be very useful as a deployment guide, but contains
sufficiently confusing statements that it might get in the way of
teaching fellow operators how to run RTR services.

Kind regards,

Job