Re: [Sidrops] [WGLC] draft-ietf-sidrops-roa-considerations-01 - Ends 10/March/2022

[Randy Bush <[e-mail]>]

why i signed on to co-author this document:

fasten seat belts, and take the little yellow pill to think at internet
scale.  the numbers below are fictitious; no prefixes or ASs were harmed
...

i have 420 /24s registered in the repo for my AS 666, folk do seem to
deaggrgate, and we are being taught not to use max-len, eh?

i move one of those /24s to my other pop/continent/whatever, AS 777

we have seen that the CA withdraws the ROA for 420 prefixes on AS 666
and then issues a new ROA for 419 prefixes on AS 666 (and one on AS777)

these two *separate* actions are published in the CA's PP

234 RPs across the internet pick them up

234 RPs across the internet chop the withdrawn ROA into 420 withdraw
VRPs and send the VRPs to their routers

ghod knows what those routers then do to BGP

the 234 RPs chop the new announcement into 419 VRPs and send to their
routers

ghod knows what those routers then do to BGP

and the poor routers are processing all that dren.  and we know the
quality of their implementations has a bit of variance

and i just wanted to move one little /24

and this whole protocol set was designed to minimize impact on routers
so it could run lightly on existing hardware

yes, i am suggesting you publish 420 separate ROAs; they're cheap.  and
then we can act on them individually.

my mommy taught me that it was not polite to 'support' a draft on which
one is a co-author.  but i do suggest others read, comment, and, if they
agree, support.

randy