[j-nsp] OSPF question

[sthaug at nethelp.no]

> I have an older M40 and an M7i. I've noticed that the M40 generates an
> intra-as LSA for it's loopback address while the M7i generates an E2 LSA.
> What's the correct behavior? How can I force the errant Juniper to generate
> the correct LSA?

A loopback address *if it is included directly as an interface* under
protocols OSPF is expected to be visible as a stub network within a
router LSA. How are your loopbacks configured?

(If your loopbacks are *redistributed* into OSPF, they will of course
appear as External LSAs.)

> Also, how do people typically deal with the differences in Administrative
> Distance values between Juniper and Cisco?

The actual numeric values aren't really that important, the relative
ordering is. Our experience is that both Juniper and Cisco have fairly
sane values for their administrative distances.

On Cisco you probably want to change your BGP admin distances such
that both EBGP and IBGP are at 200 - this is best practice and is
recommended in a number of Cisco presentations.

If you run MPLS you may need to change the admin distance for either
LDP or RSVP, depending on your choice. We ended up changing LDP on our
Junipers to have better preference than RSVP (and converting our RSVP
explicit LSPs to LDP based implicit LSPs), to have one commom protocol
for our Juniper and our Cisco routers.

Steinar Haug, Nethelp consulting, sthaug at nethelp.no