[arin-ppml] ARIN-prop-157 Section 8.3 Simplification

[Tom Vest]

On Sep 22, 2011, at 10:12 AM, Andrew Koch wrote:

> On Thu, Sep 22, 2011 at 08:57, Tom Vest <tvest at eyeconomics.com> wrote:
> 
> [snip]
> 
>> 2. Substantive objections:
>> Simplicity may be a stylistic virtue, but any benefit that might be achieved by simplifying NRPM language must be balanced against the  non-stylistic, practical consequences that the proposed simplification would have in the real world. I have explained at length, on this list and elsewhere, that I believe that subjecting ASNs to more liberalized transfer policies would have far-reaching and profoundly adverse operational consequences. I don't have time to recap the details right now, but you can find an excellent summary in the 2009 ISOC Brief entitled "IPv4 Affinity." Each of the problems that that paper identified with respect to IPv4 transfers would be vastly more problematic for ASN transfers.
>> 
>> http://www.isoc.org/internet/issues/docs/ipv6_200905.pdf
> 
> 
> While I find this summary enlightening on the topic as titled, I am
> not making the connection with ASNs.  The paper describes effects that
> LSN will have in regards to  Geolocation, Geoproximity and IPv4
> uniqueness.  Can you provide further details on how you believe these
> would extend to ASNs, which in many cases are global in scope.
> 
>> TV
> 
> Thanks,
> 
> Andrew Koch

Hi Andrew, 

I am not suggesting that Geolocation and/or Geoproximity are relevant now or might cease to be relevant for transferred ASNs, but rather that the increased mobility of ASNs could reduce the effectiveness Geolocation and/or Geoproximity for everything else. Granted the whole purpose of such mechanisms is to provide an alternative to logical/administrative topology-based distance measures, but assumptions about the stationarity of topological elements like ASNs seem to be baked into quite a few geo~ metrics and tools. See for example CAIDA's tool survey:

http://www.caida.org/projects/cybersecurity/geolocation/

For now I'll assume that the relevance of ASNs to the other four of the six issues identified in the ISOC paper ("Identifying abusers," "Spam," "Authentication and security," and "Lawful intercept/forensics") is somewhat clearer; please let me know if you disagree. To suggest just one example, an opportunity to acquire a routed IPv4 prefix bundled together with its established origin-AS on a short or long-term basis would open up whole new frontiers for strategic networking fun and games -- and if the community chooses to support market-based ASN transfers, we should fully expect that demand for such capabilities would elicit the predictable response among surplus ASN holders. Considering (just) the huge number of apparently single-homed, single prefix-originating ASNs that are currently in production, it probably wouldn't take long for the consequences of a liberalized ASN transfer policy to affect all network operators, including the vast majority that have no interest or stake in an ASN transfer policy.

As you rightly note, the ISOC paper identifies "address sharing" as the anticipated source/cause for the new challenges to come. However, the operational *effects* that the paper enumerates are not limited to things that can attributed exclusively to resource multiplexing. The most relevant passage includes the following observation:

"The important point is that today business models and commercial contracts are based on these assumptions and those business models and commercial contracts may not be sustainable as these assumptions change. Furthermore, because these assumptions are not explicit, it is possible that these changes will be made without all affected parties realizing the full impact of them in advance."

I stand by my original claim that the paper is highly relevant. 

Regards, 

TV