Skeeve Stevens wrote:
My only comment is one of possible ambiguity.
Conditions on the IPv4 address block:
- The minimum transfer size is a /24
Are we suggesting that someone that has a /22 (or whatever) can
transfer a /24 to someone out of their allocation?
Would this be a problem in the APNIC records if the /24 is in the
middleish somewhere? Is the DB prepared for:
0.0.0.0/24 Current Company 0.0.1.0/24 Current Company
0.0.2.0/24 New Company 0.0.3.0/24 Current Company
I am not sure of the structure of whois, or the support systems like
MyAPNIC to know how much of a problem this might be... I am assuming
their single allocation would be broken up into a /24 and a /23 ?
The database deals with this. Here's an example from the block I manage
via MyAPNIUC. If you look up 130.195.0.0 you
get:
inetnum: 130.195.0.0 - 130.195.255.255
netname: VUW
descr: Victoria University of Wellington
country: NZ
but if you look up 130.195.4.0 you'll see:
inetnum: 130.195.4.0 - 130.195.7.255
netname: ECS1
descr: School of Engineering and Computer Science 1
country: NZ
In this case the school is a subgroup of the university but it could say
anything there.
The better option in the database is to create three new records (/24,
/24, /23) to replace the existing /22. It's not a difficult task and
would be part of the recording of the transfer.
There are also possible small implications in routing announcements
and route pollution - the bigger the range the /24 (or something
else) is pulled from, the more pollution it will cause.
APNIC could make no promises about routability of prefixes that
have been split up in this way. That fragmentation is already common
with poor aggregation the main culprit - see Geoff Huston's Cidr Report
at http://www.cidr-report.org/as2.0/#Gains
If people did the "right thing" the global table could be reduced by
around 39%. The problem you cite is unlikely to make a significant
difference in the short term.