IPv6 in the enterprise

[Simon Leinen]

Mark Smith writes:
> sthaug at nethelp.no wrote:

>> However, we expect a significant number of (primarily) servers to
>> have static addresses and a statically configured next hop, using a
>> GUA.

> Who is in the "we" group your are talking about?

I'll join Steinar here... a significant number of servers is configured
this way today for IPv4, which will make it attractive for at least some
admins to do it that way.  I'm not saying this is "right", but people
will tend to stick with the (possibly broken) ways they have done things
for years.  And in this case I can see some benefits:

Static addresses are a reasonable way to provision servers.  You can
automate the generation of these files from a central database (which
many admins will want anyway), and you have one less boot-time
dependency on routers/DHCP servers.

Using an explicitly configured default gateway address instead of
relying on RAs is (a) how many people do it in IPv4 (anybody remember
RFC 1256?) and (b) people often prefer administratively "nailing down"
things rather than rely on a possibly insecure dynamic protocol if it
doesn't buy them much (they'll use something like HSRP/VRRP/whatever if
they want failover, will be faster anyway).

Using a global IPv6 address as the explicitly configured default gateway
address makes sense for the reasons mentioned above - the global address
is more likely to map to (and from) something sensible in e.g. DNS,
people can "ping" it more easily or traceroute to it from the other
side, etc.
-- 
Simon.