[arin-ppml] How bad is it really?

[michael.dillon at bt.com]

> I worked for a large provider and we completely automated this process.
> Our registration system would automatically send in SWIPs for all
> address
> registration transactions. It only took a few lines of PERL code to
> make
> it all work. It even parsed the ARIN responses and notified a human if
> there was a problem.

You only THINK that you automated it. SWIP transactions are emails. What
happens when some of them go astray. You system thinks it has updated
ARIN's db but it hasn't. And your db has deleted the record so that
if someone is notified of a problem, they don't know what to do. They
have never seen a SWIP transaction and there is now no way in the system
for them to delete the non-existent customer.

Maybe somebody bypassed your tool and deleted a bunch of customers 
from the DB with a maintenance script after an audit. This kind of
thing happens all the time, and nobody notices that they missed the
SWIP process until years later, someone with clue comes along.

Keeping multiple databases in sync is hard work unless you have a system
that is specifically designed for that synchronization job. ARIN's system
is not designed for this. SWIP was a crude hack, and although the new
system is a reasonably well engineered system for transaction processing,
it is *NOT* a database synchronisation system.

> > It would also provide an opportunity for providers to share best
> practice
> > tips on address auditing and perhaps develop some tools which could
> be
> > run internally on management networks to identify unused address
> blocks.
> >
> That doesn't seem like it would be all that hard under the current
> situation

To my knowledge ARIN has never invited people to work on this kind
of thing. Why is there not an ARIN sponsored project for an open
source IP address records system? In such a project you could define
a standard database schema, and build a standard module for syncing
your ISP records with ARIN's whois directory. Then other IPAM folks
could copy and adapt that syncing module.

--Michael Dillon