Re: [Roll] updating DAO caches (was Re: Something to ADD)
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Roll] updating DAO caches (was Re: Something to ADD)



> From: JP Vasseur <jvasseur at cisco.com>
> Date: Thu, 19 Nov 2009 11:36:47 +0100
> 
> On Nov 17, 2009, at 5:12 PM, Richard Kelsey wrote:
> 
> > The difficulty is that, at present, there is no way of
> > determining how much information needs to be passed up after
> > a move.  Here is the example I posted earlier.  We start
> > with this, where A is the root:
> >
> >       A
> >      / \
> >     B   C
> >     |
> >     D
> >     |
> >     E
> >
> > Now D moves to C as its parent:
> >
> >       A
> >      / \
> >     B   C
> >         |
> >         D
> >         |
> >         E
> >
> > If only A caches DAO data, then all D needs to do is send a
> > DAO after the move.  A receives it and changes its next hop
> > for D and E to be C.
> 
> Well I think that you still need the no-DAO since A may continue to send
> packets to D via both B and C.

Yes, if DAOs can be sent to multiple parents.  No if DAOs
are only sent to one parent (because then A will only
store the most recently received path to D).

> > no-DAO before the move, to update B, and both D and E need
> > to send DAOs after the move, so that C knows that it has
> > downward paths to both of them.
> 
> Note that with DAO packing only D will send one DAO for local prefixes
> and E's prefixes but that's not the topic.

No, because D doesn't have a cache.  E has to send its
prefixes, whether packing is present or not.

> > In the first case a single DAO needs to be sent.  In the
> > second case D has to send two DAOs and the nodes in its
> > subdag, just E here, each need to send one.  If only the
> > root caches DAOs then the first case always applies.  If
> > other nodes may cache DAOs, then we have to assume that we
> > are in the second case after every move.
> 
> Yes but I think that we need no-DAO to avoid nodes upper
> in the DAG to continue to use an old path anyway and the
> issues of the number of DAO is IMO larger and packing will
> sort it out.

The issue is only partly the number of DAO.  It is also who
has to send them.  If switching parents requires that every
node in the sub-DAG send a DAO, we need to take that cost
into account and avoid it when it is possible to do so.  DAO
packing is not a magic bullet that makes DAOs free.

                                 -Richard Kelsey

Note: Messages sent to this list are the opinions of the senders and do not imply endorsement by the IETF.