You are here: Home > Participate > Join a Discussion > Mailman Archives
<<< Chronological >>> Author Index    Subject Index <<< Threads >>>

Re: Changes to PI Policy?

  • To: leo vegoda < >
  • From: Sabri Berisha < >
  • Date: Tue, 15 Apr 2003 22:39:03 +0200

On Tue, Apr 15, 2003 at 08:17:25PM +0200, leo vegoda wrote:

> If the concept of PI is not broken then the most obvious question
> is, should there be a minimum assignment size for PI?
> If those requiring PI received a minimum prefix length (and then
> whatever they need) then we could be sure the addresses were useful.
> However, this would probably mean that many people that could
> genuinely use a /28 would be be eligible for more addresses,
> probably a /24.

A /24 is generally considered to be the smallest network still routable. I
know of no tier-1 who announces a prefix longer then a /24 to their
peers. However, sometimes that could be exactly what you want. An
example of this is which is used by a customer of
mine who uses this space to make sure a network is only reachable on a
limited part of the internet.

So, yes and know. If the PI-requester needs to have global routability,
a /24 would be the smallest block that could be assigned.
> If the "Land Rush" is considered to be a dangerous possibility then
> it might be possible to introduce qualifying criteria. At this point
> it gets tricky as it's necessary to start chosing who is 'special'
> and who is unlucky. For instance, a root DNS server might well be
> special, a ccTLD DNS server? an IXP (yes, I know there are good
> technical reasons for not using PA). I assume that my home network
> is not very special, but some people might thing that their SOHO 
> networks are critical infrastructure.

I have received a /24 PI space for services which are DDoS sensitive
(yes, that includes, but is not limited to IRC). This has saved my
customers and me a lot of downtime so I would definately say that this
should be one of the criteria to assign PI space. So far I was not able
to detect any loss of connectivity due to it being a small PI block. 
> Finally, it is necessary to consider that the space in 193/8 and
> 194/7 is a finite resource. In the two years between October 2000
> and October 2002 we assigned just under a /13 and a /15 of PI (in
> total[1]). The total size of the space we have left in 193/8 and
> /194/7 is just over a /13. We do not make PA allocations from these
> /8s any more, to make sure that we have PI space available for the
> forseeable future.

Will there be more space available for PI assignments (as in, will IANA
allocate more to the RIPE NCC)?

Sabri Berisha   "I route, therefore you are"

Per user RBL checking:

Attachment: pgp00000.pgp
Description: PGP signature

  • Post To The List:
<<< Chronological >>> Author    Subject <<< Threads >>>