[db-wg] out of region routing in the RIPE Database

denis walker via db-wg wrote:
> We would like to start off the (final?) round of discussions on this
> topic with a question. Should the RIPE Database be used as an open,
> free, global Internet Routing Registry (IRR)? If the answer is 'yes'
> then perhaps we should allow the routing policy of any global resource
> to be documented in the RIPE Database as a choice by the resource holder
> and move on to address the issues surrounding this process. If the
> answer is 'no' then perhaps we shouldn't allow any routing policy for
> any non RIPE resources (not selectively try to expel one group of ROUTE
> objects).

I am not in favour of having the RIPE database as an open-access
database on the basis that this mixes up two sets of data, authoritative
and non-authoritative, and it it is impossible for someone casually
querying the database to determine which is which.

Some people are inserting random route: objects into the database, and
those route: objects are being picked up by provisioning systems and
ending up configured on routers and IXP route servers.  This enables
prefix hijacking, which is a pressing operational issue.

There are two main ways to fix this problem:

1. change all non-RIPE address space to have a different source: tag

2. remove all non-RIPE address space completely

There is some previous discussion about this issue on db-wg, and IIRC,
the WG suggestion was to go for option #2.

On that basis, again IIRC, there was a suggestion to handle this in a
phased way, starting off with the lowest hanging fruit first.  As
Afrinic objects formed the largest share of all out-of-region objects in
the ripedb, they were targeted first.  I believe the plan was to
continue on after that with other categories of objects. This
recollection may be wrong.

Nick