[arin-ppml] Rationale for /22

Ted Mittelstaedt tedm at ipinc.net
Tue Jul 28 14:31:28 EDT 2009

Chris Grundemann wrote:
> On Tue, Jul 28, 2009 at 11:02, Kevin Kargel<kkargel at polartel.com> wrote:
>>>   Obviously the primary reason for a limit is the accepted minimum routing
>>> entry size of a /24.  Of course, you knew that.  So why ask the question?
>> Perhaps it is time to examine the 'accepted minimum routing entry size'.  If
>> it won't cause problems then why not extend the mask size?  Or are you
>> saying that long mask route entries do in fact cause problems?
> I don't think that anyone said lowering the minimum routing entry size
> would not cause problems.  I am fairly certain that everyone agrees
> that <rapid> route table expansion is a potential problem (primarily
> caused by long route entries due to de-aggregation) and that we need
> to continue to help ensure that the route table does not grow
> explosively.
> The distinction that I (and I believe others) make is between the
> source of a given Org's space and the addition of entries in the
> table.
> AFAIK, there are no ISPs allowing customers to multihome with less
> than a /24 currently but many who are allowing those /24s.  So if that
> Org gets a /24 from ARIN or a /24 from it's ISP, there is the same 1
> entry in the table.  And when that Org needs another /24 (and the ISP
> did not reserve the adjacent one and the customer does not want to
> renumber) and the ISP hands it another one, or ARIN hands it another
> one, there are the same two entries.  There are two main differences
> here though:
> 1) If ARIN policy requires that any Org with a /24 that wants more
> space be required to return the original /24, then we end up with one
> /23 instead of two /24s in the table.  This is more likely to happen
> than most ISPs requiring renumbering because of all those pesky sales
> teams and account managers...
> 2) If ARIN is giving space to a greater portion of all multihomers,
> then ISPs have less excuses not to better aggregate their own space.
> Load balancing should never require de-aggregation down to /24s.
> Unless you stop allowing Orgs to multihome with less than a /22, there
> will be longer entries in the table.  Perhaps we could actually slow
> routing table growth by allowing them to get there space from ARIN
> (since they will be getting and announcing it anyway)   Or do you
> think that getting space from ARIN will be so much easier than getting
> it from an ISP that Orgs will flock to /24s and cause more growth in
> the table than we currently see?

If they get their /24 from an ISP and advertise it, that likley will be 
part of a supernet that that ISP is advertising so if I want to filter
to the /23 level I won't be blocking off reachability for myself - since
presumably the LIR will know how to reach it's own customers.

But if they go directly to ARIN and get their /24 then I'm going to have
to filter at /24 boundary.


More information about the ARIN-PPML mailing list