Taking Carlos' analysis a step further, there are two kinds of "stuff" that has to be fixed during a renumbering event: stuff that can be fixed automatically (cost essentially independent of network complexity) and stuff that has to be fixed manually (cost that scales with network complexity). For example, addresses assigned through autoconf can be fixed automatically while hard-wired addresses buried in config files have to be fixed manually. At the risk of over-simplification, there will be cost savings in renumbering IPv6 networks - cost savings that can be directly attributed to IPv6 - if there is more stuff in the first category for an IPv6 network than for an IPv4 network. That is, if we can point at stuff that can be fixed automatically in an IPv6 network, we can identify cost savings.
The cost savings analysis in the previous paragraph focuses on time (and $$$s) saved by network admins. Another cost metric might be time-to-renumber; that is, how quickly can a renumbering event be completed. For example, IPv4 hosts using DHCP can be automatically renumbered, but there is a latency in the renumbering related to the length of the leases on the addresses assigned through DHCP. IPv6 hosts using autoconf can be automatically renumbered by changing the advertised prefixes, presumably with less latency.
Another potential advantage for IPv6 that is a little harder to quantify is the notion of "graceful" renumbering - the ability to have a transition state in which both the old and new prefixes are in use simultaneously. Will this transition state work in practice? Has anyone actually tried it?
So, here are three ways to assess the renumbering advantages for IPv6:
* Stuff moves from "manually configured" in IPv4 to "automatically configured" in IPv6, as a direct result of the design of IPv6 * Renumbering latency is reduced * Renumbering is enabled by IPv6 (graceful renumbering) that can't be accomplished with IPv4
- Ralph
At 11:24 AM 8/14/2003 +0100, Tim Chown wrote:
Carlos, the question is "where are the cost savings in renumbering an IPv6 network as opposed to renumbering an IPv4 network".
Tim
PS. According to dnsop people, everyone will use DHCPv6(!)
On Thu, Aug 14, 2003 at 11:20:07AM +0100, Carlos Friacas wrote:
> On Thu, 14 Aug 2003, Tim Chown wrote:
>
> > On Wed, Aug 13, 2003 at 02:57:05PM -0400, Keith Moore wrote:
> > >
> > > IPv6 gives users better ways to solve some problems (renumbering,
> > > attachment of a home network) whereas in IPv4 NATs were the
> > > best tools available.
> >
> > So where are the cost savings in renumbering an IPv6 network as opposed to
> > renumbering an IPv4 network? This would be an interesting question to expand
> > on I think...
>
> renumbering an IPv4 network:
> - The 1st critical parameter is *size*.
> *if* you are not using dhcp, even in an /27 segment (30 hosts if 100%
> used) it can be real hard to renumber. it consumes *time*, although being
> a simple/basic task. If you have a /24 or /23, well.. it is a real pain.
> - The 2nd critical parameter is how many "hard-wired" addresses
> (regarding services) you have inside it.
> dns servers are usually the biggest pain. if you use dhcp, resolvers are
> not a problem (if they are inside the network being renumbered), you just
> have to change a parameter in the dhcp server, if not it is a pain.
> but the trickiest/hardest part usually lies with authoritative
> nameservers -- If your "upper levels" (forwards and reverses) aren't fast
> enough to respond, you will think twice before going into a renumber. :-)
>
> renumbering an IPv6 network:
> - The 2nd critical parameter i've mentioned is also valid for IPv6. :-(((
>
> One more comment:
> Renumbering from 6BONE addresses to RIR addresses... it might be a good
> solution/idea to still route the /128s associated with addresses that are
> in somewhat way hard-wired to some services, as part of the phase-out
> process. e.g: keep the routing to 3FFE:XYZW:1:2::1/128 in your IGP to
> 2001:XYZW:1:2::1/128, but stop configuring 3FFE:XYZW:1:2::/64 on its
> gateway/router. This will prevent other machines to get a prefix in 3FFE,
> but still guarantees the connectivity to one single server in the "6bone
> context".
>
>
> Regards,
>
> ./Carlos "Upgrade the Internet! -- Now!"
> -------------- [http://www.ip6.fccn.pt] http://www.fccn.pt
> <[EMAIL PROTECTED]>, CMF8-RIPE, CF596-ARIN, Wide Area Network Workgroup
> FCCN - Fundacao para a Computacao Cientifica Nacional fax:+351 218472167
>
> "Internet is just routes (125953/461), naming (millions) and... people!"
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
-------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
