Re: [Ospf-wireless-design] Advantages of aligning adjacencies with relays

[Richard Ogier <[e-mail]>]

I would like to go into more depth regarding the following advantage
of having a set of relays (MDRs) that are responsible for forwarding
all LSAs (independent of the originator) and for maintaining
synchronization with neighbors.

 > Extendability to non-ackable LSAs (i.e., periodic flooding).
 > By having a fixed set of nodes responsible for flooding LSAs
 > (independently of the originator) and maintaining
 > synchronization with neighbors, it is easier to implement
 > options such as non-ackable LSAs (described in Appendix D of
 > the MDR draft), by having each MDR flood each LSA periodically
 > until a new instance has been received.

One might argue that MPRs can be used for periodic flooding
of LSAs, as in OLSR.  But let's take a closer look at this
in the context of OSPF. To accomplish periodic flooding using
MPRs, the originator (advertising router) must periodically
flood its router-LSA (e.g., every 5 seconds) whether or not
the LSA changes.  Let's first suppose that the LSA sequence number
is incremented only when the LSA changes.  Then an MPR that
receives a periodically flooded LSA from an MPR selector, which
is not a new LSA (has the same sequence number as the database copy),
must still flood the LSA for this to work.  But this is contrary to
OSPF, in which a received LSA is forwarded only if it is a new
instance, and it is not clear what the forwarding rule should be
if the same MPR receives the same LSA from another MPR selector
1 second later, or 5 seconds later, etc.

Now suppose that the LSA sequence number is incremented (by
the originator) with each periodic flood, whether or not
the LSA has changed.  This is also contrary to OSPF, and will
cause unnecessary flooding of the LSA over non-MANET
interfaces that are in the same OSPF area.

Now consider the periodic flooding method (using non-ackable LSAs)
proposed in the MDR draft.  As in OSPF, the originator increments the
LSA sequence number only when the LSA actually changes.
As in OSPF, a received LSA is forwarded only if it is a new instance.
The only difference regards local retransmissions: each MDR periodically
multicasts each LSA until a new instance of the LSA is received.
(If topology changes are frequent, the new instance will often be
received before the old instance is multicast a second time.)

This method does not work with MPRs because each MPR is responsible
for forwarding only LSAs received from its MPR selectors, and
it is not clear what to do when the set of MPR selectors changes
without receiving a new instance of the LSA.

This is just one example of how the use of source-independent MDRs
makes the design more extensible (and more compatible with OSPF).

I will mention here another advantage of the fact that OSPF-MDR
treats a MANET as a generalized OSPF broadcast network.
This allows OSPF-MDR to treat an OSPF broadcast network as
a special case of a MANET, at least for the purpose of flooding
LSAs between a broadcast interface and a MANET interface.
This is described in Section 8.1.2 of the MDR draft.

Richard




_______________________________________________
Ospf-wireless-design mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/ospf-wireless-design