[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
IPv6 routing issues
> As of the last time I talked with the IDRP document authors, they were
>indicating that they did not consider their draft fully cooked. Until they
>make sufficient progress refining their draft, it seems unwise to depend on
Hmmm. Mabye we need new authors in fact I am working off line now to
find some authors. I was unaware anyone was actually working on IDRPv6?
> Similarly, my (possibly outdated) understanding has been that the OSPFv3 spec
>was still evolving. For example, there were some unresolved discussions about
>whether the OSPFv3 spec could be changed to permit OSPFv3 to route both IPv4
>and IPv6 (analagous to how Integrated ISIS works). The feature of "integrated
>routing" is one that our customers are clearly asking for. Many customer
>sites view integrated routing as critical to their ability to transition from
>IPv4-only to IPv4+IPv6 routing.
Well we have a pretty good OSPFv6 spec out there now and I am hoping Rob
Coltun does an implementation. So I am not concerned if it later has to
be updated to v3.
As far as I-IS-IS that is a nice to have but IPv4 and IPv6 routing is a
bit different than tunneling IP in CLNP or the other way around.
Besides Dimitry's routing-aspects draft covers this anyway so my gut
feeling is I-IS-IS is not needed for IPv4/IPv6 and knowing it well I
would say this is just a matter of explaining this to customers as
education. As most likely the ones who want it are using I-IS-IS today
and thats not a lot of customers really. Bottom line is you get
integrated routing with IPv6 for IPv4 automatically by definition of how
the architecture was designed for IPv6 and the relative specs.
I am also hearing from Friends there is a draft on TAGS to use up part
of the flow-label in IPv6 and I innately have a real problem with that
and will be bringing that to the IPng WG list soon.
> RIP can function fine as an exterior routing protocol in many environments,
>so lack of IDRP or OSPF is not necessarily an operational issue. Depending
>on how well RIPng converges, network service providers might find it desirable
>to work on thoughtful provisioning but that is an operational issues for such
>providers to sort out among themselves if/when such providers exist.
True... But IDRPv6 will provide a lot more to the telcos and ISPs.
> Similarly, I'm hearing major commercial customers express dismay that IPv6
>does not explicitly include support for EIDs.
I speak with a lot of customers and none have brought it up other than
in terms of DNS or LDAP to locate people not nodes on a network. In a
tutorial I do and have done for two customers I go over the history of
how all this came about. Both customers state what I just stated on
IPng and what Steve just responded, and that is but they cannot be
defined and I need more than one of them.