The MPLS-OPS Archive

[George Matey <gmatey@equipecom.com>]

>  > > 
>  > > what is the recommended IGP for an MPLS TE using RSVP? 
> ISIS or OSPF?
>  > > 
>  > The only difference is that for TE updates, ISIS uses different 
>  > different TLVs for each update, whereas OSPF may include 
> more info in a 
>  > single TLV object. 
> 
> ISIS is somewhat more efficient in the way it uses the space in the
> packet - multiple TLVs of different types can be put into the same
> packet.  With OSPF, in general, each different LSA type uses a
> separate packet.
> 
> ISIS also has partial recomputations of the topology for certain types
> of minor failuers.  OSPF must do a full recalculation of the topology.
> 
> ISIS appears to be a bit more scalable than OSPF.  There's at least
> one ISIS deployment with over 1000 routers in the domain.  The largest
> OSPF deployments seem to max out at roughly half that.


I've heard similar stories from various mailing lists, but just to report
a larger "upper bound" for OSPF scalability:  MCI had a network where
they had approximately 1300 routers in a single area before they were
forced to OSPF areas.  Obviously a lot of design and development work went
into that network, and it was the largest single-area OSPF deployment
I'm aware of.  I'm not saying that OSPF is better or worse, just that most
people's perceived "upper bound" for OSPF tends to be low.

> 
> I don't think it matters all that much which protocol is used.
> 
> Daniel


For large network topologies, the design of the network is just as
important as the implementation of the protocol.  A great protocol
implementation can't necessarily scale a poorly designed network.

--
George

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml