|
For implementations that follow the RFC 2362 word-by-word (not sure if
there are any::) where one needs to keep the stats for all the packets
received and switches to SPT only if packet count is above threshold, Don's
argument makes sense.
I guess one thing that could be done in such case is if R2 wants to
join the group and R1 doen't have any other oifs, then we might use a version of
assert to make R2 the DR for the time being. If a receiver comes and gets
attached to R1 at any point of time, let R1 take over as DR again.
Just a thought.....
-Riyaz
----- Original Message -----
Sent: Thursday, March 03, 2005 5:34
PM
Subject: [pim] (no subject)
I think the one extra hop count you meant will happen for only
one packet/time. As soon as the Receiver receives the first packet from the
Source, it will break the RPT and forms SPT(PIM-SM). From next packet onwards
mcast traffic will follow the shortest path. Definitely we should have DR for p2p links too, so
that the multicast link is under one router
control.
Thanks
Ganesh.
Isidor,
Thanks for the quick
reply. Thinking about it a little more, consider
the
following.
Suppose R1 and R2 are on the ends of the p2p link, and R1
is the DR, and
R2 wants to join a group on the interface on that link.
This would mean
that R1 would send the join, not R2. If R1 and R2 both
have p2p links
to R3, and R3 is the way to get to the RP (PIM-SM) or the
source
(PIM-SSM), then the join would go from R1 to R3. It all works, but
when
multicast traffic comes back there is an extra hop: R3 to R1 to R2.
If R2 had been the DR, then would it not send the join to R3? In
this
case, there would be one less hop, compared to the above. It seems
that
if a router wanted to join a group, then the most efficient result
would
be obtained if that router itself sends the join, even in more
general
cases than the one described.
I'm not sure if this is the
correct interpretation of what should occur.
In any case, your thoughts
would be welcome.
Thanks,
Don
-----Original
Message-----
From: Isidor Kouvelas [mailto:kouvelas at cisco.com]
Sent:
Tuesday, March 01, 2005 4:23 PM
To: Smith, Don
Cc:
fenner at research.att.com; M.Handley at cs.ucl.ac.uk;
holbrook at cisco.com;
pim at ietf.org
Subject: Re: DR election
One
of the two routers may join a multicast group on its p2p link
interface.
The DR on that link is responsible for initiating the
relavant
joins.
Isidor
Smith, Don wrote:
> PIM Folks,
>
> In draft-ietf-pim-sm-v2-new-11.txt, Protocol Independent Multicast
> -Sparse Mode (PIM-SM):Protocol Specification (Revised), there is a
> comment
>
> We note that some PIM implementations do
not send Hello
> messages on point-to-point interfaces, and so cannot
perform
> DR election on such interfaces. This in
non-compliant
> behavior. DR election MUST be performed on ALL active
PIM-SM
> interfaces.
>
>
> Please let me know why
DR election is useful even on point-to-point
> links. The reason for
Hellos is clear, but not DR election. Perhaps
> I'm missing an
important point.
>
> Thanks,
> Don Smith
>
_______________________________________________
pim mailing
list
pim at ietf.org
https://www1.ietf.org/mailman/listinfo/pim
Thanks and Regards,
Ganeshbabu
V
__________________________________________________________________________________________________
"Tomorrow's life is too late. Live today."
_______________________________________________
pim mailing
list
pim at ietf.org
https://www1.ietf.org/mailman/listinfo/pim
|
