apnic - day three

[Randy Bush <randy@psg.com>]

the net is about 90% usable today.  the rate of laptop infection is
down to one every hour or two.  there is a team watching for them,
their external access is being blocked at the border, and the help
desk is cleaning them up for the lusers.

itojun and i met with woohyong choi who is a key tech clue for the
seoul ietf net team.  he seemed good, and assured me that it would
be a very different network provider, kt, though there might be
issues with kt's intl connectivity.  we agreed that there will have
to be an entire net security team for the meeting.

more address policy:

  o draft-hinden-ipv6-global-local-addr-02.txt discussed and
    discussion of rir response.  the rirs could do it if the ietf
    is really this warped.

  o paul wilson presented the proposal that the hd ratio, as
    opposed to 80%, be used as the trigger for a new allocation in
    ipv4.  it may be biased toward larger providers, but is not a
    major policy change that i can see.

  o geoff huston continued frank's solensky's classic ipv4
    utilization lifetime measurement report with current data and
    using a three-fold (iana, rir, and routing table) analysis.

    looking at iana /8 allocation, solensky's 2008 now looks like
    2019.  looking at rir allocation rates, it's 2028/9.  looking
    at the routing tables for what is actually announced, it is
    2028/9.

database sig, see presentations at
<http://www.apnic.net/meetings/16/programme/sigs/db.html>:

  o a presentation suggesting that inter-irr database transfer be
    scopable using an unspecified protocol.  i.e., if 
    - ripe mirrors arin, and 
    - lacnic mirrors ripe, then
    - arin might make it so that lacnic does not get arin's data
      from ripe
    i wondered if all this mirroring was silly and merely trying
    to compensate for whois being broken and not doing a full
    search.  the discussion then wandered in confusion.  whois
    seems to do this to people.

  o discussions of processes apnic is doing on privacy of data and
    cleaning of old data.

dns operations sig, presentations at
<http://www.apnic.net/meetings/16/programme/sigs/dns.html>:

  o george michaelson discussed proposed procedure changes to
    handle the problems caused by lame servers for reverse
    delegations.  proposed process
    - identifying potential lameness for 15 days
    - 45 day notice period
    - disable delegation

    to be noted that there is an acceptance of single server
    delegations in the region.  bad bad.

  o i presented the same old stuff on delegation of
    2.0.0.2.ip6.arpa

  o geoff presented some hacks based on keith's draft, but did not
    suggest them.

routing sig, presentations at
<http://www.apnic.net/meetings/16/programme/sigs/routing.html>:

  o geoff huston presented measurements of routing announcements of
    prefixes and ASns which were actually not allocated, i.e. are
    bogons.

  o geoff presented his weekly report of counts and sizes of ipv6
    prefix announcements.

  o nagahashi kengo presented a survey of how much the registry
    routing database is actually used.  this was essentially on how
    horribly inaccurate the data in the irrs is.  ripe's routing
    data are better than the others.  the speaker felt they were
    good enough to actually use.

  o masasi eto presented on improving the reliability of the irr.

  o i presented some measurements on how horribly noisy bgp
    actually is.

randy