On Thu, Nov 19, 1998 at 04:59:01PM +1300, Craig Anderson wrote:

Agreed. Since Iprolink will not have a representative attending the ISOCNZ conference. I will state our company's position on ownership:

Speaking only for myself;

1) ownership of NZGATE addresses belong to the company to which they were assigned. In the case of ISPs sub-assigning blocks, this may be a contractual matter between the ISP and the client;

So originally these addresses belonged to NZGATE and were transfered to the companies with which they now reside? NZGATE/Waikato has a little peice of paper which said they owned these addresses?

3) we would view attempts by a group of NZ ISPs to block or otherwise restrict portability of the NZGATE addresses as illegal.

Illegal - which law or laws would be in violation here?

1) in 1995 there was very valid concern that because routing table growth was outstriping router CPU and memory growth that is was only a matter of time before major routing problems developed;

2) the introduction of CIDR, minimum allocations sizes (no new assignments longer than /19), and improvements in routing software, have slowed router table memory and CPU requirements to much less than improvements in technology are delivering and can be expected to deliver. In short, the changes have removed any _need_ to introduce further changes and in fact indicate a relaxing of rules is more likely;

Did 2 occur after 1? CIDR predates 1995 (see rfc1518 and rfc1519). In fact, drafts and discussions about it go back to 1992 or earlier I would guess. Which 'improvements in routing software' are we talking about here? The size of the routing table for machines is less of an issue as time progresses (providing the growth of the 'net doesn't outstrip Moore's law) -- but more routes is going to lead to more human errors and more complicated access-lists, etc.

3) The only route filtering occurring on the Internet today that affects connectivity (i.e. backbones with no default route) is being done on /8's where the minimum allocation size is /19. With half the root nameservers on /24's, wholesale short prefix filtering will not occur in the short term. Attempts to predict long term policy are not valid.

Filtering can be done on a per object basis. People could quite feasibly refuse to accept routes with excessively long prefixes for any designated network or networks if they choose to do so -- this need not impact upon existing networks or indeed the root name servers.

4) The NZGATE addresses are in a /8 block where the minimum allocation size was /24;

Is this `just how it is' or shaped by policy -- if the later, can someone please tell me where I mind find this. Was 202/8 initially not divided up into much large chunks (eg. /20)?

5) With 40,000+ routes, it's hard to make a case that the addition of several hundred routes due to migration is significant (more than several hundred additional routes are unlikely because of aggregation and the likelyhood of distribution tending to favor a few ISPs).

Indeed such a small increase isn't an issue -- so long as it remains only a small increase. If we decided all networks are portable regardless of size, then over time severe fragmantation could occur.

1) Iprolink would support a public statement which encourages migration to new addresses (by explaining the benefits) where it would not tend increase use of address space.

This very sounds reasonable. Nobody should be forced to renumber (unless there network is very small, say /28), but there should be mechanisms which encourage this when appropriate. As someone else suggested, perhaps when they apply for more network space, they may first have to agree to relinquish some of their existing space in order to keep aggregation manageable. -cw --------- To unsubscribe from nznog, send email to majordomo(a)list.waikato.ac.nz where the body of your message reads: unsubscribe nznog