[arin-ppml] Draft Policy ARIN-2024-5: Rewrite of NRPM Section 4.4 Micro-Allocation

[Owen DeLong]

Again, I think (and if I were involved in Open-IX would argue there) that their standard is over-specified and over-constrained. While 802.3z and 802.3ae are very common interfaces today, I know, for example, that there are at least a couple of IXPs that are considering (if not implemented) the elimination of 802.3z and moved up to 802.3ae as a minimum IX connection. I think the days of every IX offering 1Gpbs connections are certainly numbered as 10G becomes ever cheaper to implement.

Yes, ARIN should take due notice of standards issued by other standards bodies, but it should not hold itself arbitrarily beholden to them if there is good reason to diverge.

In this case, IMHO, ethernet is a good example here. There’s no advantage to ARIN policy by adding this constraint. It does not in any way improve ARIN’s ability to offer service to its members and it creates a situation where technological advances have a relatively high likelihood of rendering policy obsolete faster than it can be updated.

That doesn’t seem like a good thing IMHO and unless you can offer a justification for it beyond “It’s what Open-IX says”, I think there is good reason to choose alternative language for this policy.

Owen


> On May 28, 2024, at 22:03, Martin Hannigan <hannigan at gmail.com> wrote:
> 
> 
> I wanted to add a little more color on the ethernet language in the 4.4 revision proposal. I read somewhere earlier today someone well-known pointing out the importance of consensus based standards. The IETF is a consensus based SB. And according to their ANSI accreditations so are IEEE (electronics), TIA (telecom), NEMA(electric, fiber optic enclosures, et. al.), AQM (ISO 9001), and Open-IX.
> 
> ARIN, IMHO, should use the existence of a bonafide standard (and SB) as a strong and meaningful input with regards to meeting principles of being technically sound.
> 
> -M<
> 
> 
> Physical Interface
> 
> The IXP MUST be able to offer IEEE 802.3 Ethernet connectivity on a common
> switch infrastructure. Service offerings MUST be available at least at the
> following IEEE defined rates:
> 
> - 802.3z 1GE
> - 802.3ae 10GE
> 
> [ clip ]
> 
> Traffic Forwarding
> 
> The IXP MUST forward frames with the following Ethertypes:
> 
> 0x0800 IPv4
> 0x86dd IPv6
> 
> [ clip ]
> 
> 
> 
> On Sat, May 25, 2024 at 12:46 AM Martin Hannigan <hannigan at gmail.com <mailto:hannigan at gmail.com>> wrote:
>> 
>> 
>> On Fri, May 24, 2024 at 4:42 PM Owen DeLong via ARIN-PPML <arin-ppml at arin.net <mailto:arin-ppml at arin.net>> wrote:
>>> I think the best resolution here would be to move all definitional text for IX in both proposals into a proposed (and identical) definition in NRPM section 2 and then it’s harmonious and not in need of future synchronization regardless of whether either or both proposals are adopted.
>> 
>> I do agree with Bill's suggestion that we finish 2024-4 including a definition and defined term for 2024-5 and anything else related ("Defined Term"). So I think we're agreeing.
>> 
>> I object to the removal of "ethernet" from 2024-5.
>> 
>> Warm regards,
>> 
>> -M<
>> 
>> 
>> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20240528/f3d27f85/attachment.htm>