[ppml] HD Ratio Applied to IPv4
pwilson at apnic.net
Wed Aug 13 21:03:28 EDT 2003
For background on the HD ratio, including the justification of this approach
in preference to percentage-based utilisation measures, please see RFC3194.
> -----Original Message-----
> From: owner-ppml at arin.net [mailto:owner-ppml at arin.net] On
> Behalf Of Charles Scott
> Sent: Thursday, 14 August 2003 7:02 AM
> To: sigma at smx.pair.com
> Cc: ppml at arin.net
> Subject: Re: [ppml] HD Ratio Applied to IPv4
> I'm sorry, I'm just not getting this. It would seem that
> any particular level of assignment needs to stand on it's
> own. Why then would more levels of assignment below a
> particular level make that one level more challenging to manage.
> Also, the larger the assignment, the larger 20% of the
> assignment becomes, so it would seem to be easier to manage
> larger size blocks unless the size of the sub-assignments for
> some reason become proportionately
> lager relative to the size of the total block as one manages
> larger and
> larger blocks.
> I don't say the above rhetorically. If there's a good
> reason to make this kind of adjustment then it should be
> done. But when we're looking at a potential for significantly
> large additions of wasted (unassigned) space added to the
> equation, there should be darn'd good reason for making the
> change, particularly when it adds an additional level of
> confusion to the process.
> On Tue, 12 Aug 2003 sigma at smx.pair.com wrote:
> > > APNIC is looking at applying the HD Ratio system to IPv4
> > > allocations.
> > > Attached is their draft document discussing the idea.
> I'm curious what
> > > people in the ARIN region think of the idea.
> > The current situation, where 80% utilization is required
> for any size
> > block, is more challenging when you have multiple levels of
> > downstream, subdividing the block, wasting boundary
> addresses, making
> > it difficult to ensure that each assignment is well-utilized, etc.
> > The AD-Ratio idea seems to reverse the situation. Now things are
> > smooth for networks with multiple levels of assignment, but
> waaaay too
> > loose for simpler networks. A company which has a /17, for
> > and uses it entirely for their own networking, can now get new
> > allocations after only 70.22% utilization. These companies (eg,
> > datacenters, Web hosts, large workstation clusters, etc)
> have the best
> > opportunity for utilization nearing 99%. This effectively
> gives those
> > companies less reason to be efficient.
> > Is there any way to distinguish the two scenarios usefully
> and adjust
> > the idea accordingly?
> > Kevin
More information about the ARIN-PPML