In a message written on Thu, Dec 10, 2009 at 01:35:16PM -0500, Jared Mauch wrote:
As always, good research by renesys.
http://www.renesys.com/blog/2009/12/bonjour-yall-asn-split-persona.shtml
As already commented on the blog... ISC had a data entry error on an ASN for our site in Fiji. There was no RIR mixup, it was purely an error on our part. This was then further propogated when scripts we had generated routing registry objects and pushed them out. We're already down the path of fixing it, and the error will be corrected soon. We would like to thank Renesys for bringing it to our attention. While the ARIN / RIPE mixup in the 17xx range has caused a lot of people to go looking for a smoking gun I think Renesys has proved there is in fact no cause for alarm. 40,000+ ASN's in use and only two for which there is even a question. 0.005's of a percent error rate in a global system is quite good. To also answer one of the questions posed in the blog. It is only recently (I think about 4 months ago) ISC fully scripted it's routing registry updates, which is what caused the AS35686 to ISC entry in the RIPE DB. Prior to that there would have been no ISC entry anywhere, as it was not assigned to ISC; thus no other party would have checked it. I can't comment on if ISC checked if the ASN was in the APNIC database properly when they received it or not, but noting it was a data entry typo it is entirely likely the database was checked with the proper ASN, and then the data was miss-entered into internal systems. I would be very interested to know if something similar happened with AS3745. -- Leo Bicknell - bicknell@ufp.org - CCIE 3440 PGP keys at http://www.ufp.org/~bicknell/