[arin-ppml] New Policy Proposal

[John Curran]

On Aug 16, 2012, at 10:59 AM, Bill Sandiford <bill at telnetcommunications.com>
 wrote:

> John:
> 
> The CRTC CISC process that you have mentioned is actually a totally separate process.  That process, which I was part of, was a process to come up with a solution for Third Party ISPs to be able to assign static IP addresses over the cable companies infrastructure.  To date, the only available assignment method is, and remains, dynamic allocation via DHCP.

That doesn't match how the problem was explained to me, but I defer to 
your experience in this area.  It does, however, raise an question which
inline below.

> Bill:
> 
> The problem that this policy attempts to address is a result of the framework that the CRTC has mandated for the Cable Companies to open up (or unbundle) their internet access services in a similar fashion to what the ILECs are also required to do here with DSL.
> 
> The CRTC passed a framework that mandated the cable companies to provide third party internet access (TPIA) to any competitive ISP that chose to do so.  The problem is that the CRTC did not define specifically as to how it would work technically and left those decisions to the cable company (ie the CRTC didn't want to tell the cable companies how to run their network, they just mandated that it be open).
> 
> When a TPIA ISP subscribes to the cable companies wholesales services the cable company REQUIRES that the third party ISP provide an assignment of addresses large enough so that they can adequately provision all of the CMTS's in their network for the service for assignment via DHCP.  This is usually done by reserving a /29 or /28 for each node on a CMTS.  To be clear, the TPIA ISP has no control over this process.  If they want the service they must provide the required amount of IPs to the cable co.  No IPs, no service.  This has typically required a /18 or /17 of space to accomplish.
> 
> Most ISPs have been able to acquire this *initial* space by making use of the immediate needs policy.  The problem occurs however when the ISP needs additional space.  For example, when one node (or two or three) fill up and exhaust the /29 or /28 that was assigned to them they require an additional block for that node.  In some cases additional blocks are required when the cable company augments their network and does a "node split" which requires another /28 or /29 to get used.  The ISP then makes a request to ARIN for additional space however they are unable to receive additional space because, although the address space is assigned to equipment, it doesn't meet the current definition of "utilized" under policy.

The ISP's entire block does not meet the minimum utilization criteria,
since it is only the allocation for one node (or two or three) that have 
filled up.

Given that addresses from these per-node pools are dynamically allocated 
to subscribers, is it possible for the cable company to reprovision the 
address space from the ISP's other underutilized nodes as needed?  (This 
is what is expected from all other ISPs in similar situations under the
present policy.)

Thanks!
/John

John Curran
President and CEO
ARIN