Hello Vicky,
This all looks good to me...
Regards,
Charlie P.
On 4/30/2015 8:23 AM, Victoria Mercieca wrote:
Hi Charlie,
I've added this in the Message Transmission section:
To avoid congestion, each AODVv2 router's rate of message generation SHOULD be limited. The
implementation is free to choose the algorithm for limiting messages, including prioritizing
messages when approaching the limit, and the rate (CONTROL_TRAFFIC_LIMIT) should be administratively
configurable. AODVv2 messages SHOULD be discarded in the following order: RERR for invalidated
routes, RREQ, RREP, RERR for undeliverable packet, RREP_Ack.
So it hints at the points I made below without actually specifying the behaviour. I hope that's not too lazy of me! If we get to that limit, route maintenance is going to be affected anyway, but this enables a slightly more graceful way to ease the situation it before it reaches the limit.
Regards,
Vicky.
On Thu, Apr 16, 2015 at 10:56 PM, Charlie Perkins <charles.perkins@xxxxxxxxxxxxx <mailto:charles.perkins@xxxxxxxxxxxxx>> wrote:
Hello folks,
More follow-up below...
On 4/16/2015 7:02 AM, Victoria Mercieca wrote:
This email contains questions about the control message limitDropping is simplest. I'd be O.K. if you wanted to make more text
(Section 9 in Version 9a or Section 6.3 in Version 9b):
* If the limit is reached, should any AODVv2 messages we would
have sent get buffered? Draft currently says they would be
discarded. RREP, RREP_Ack, and RERR for an undeliverable
packet seem too important to just drop.
to refine this however.
No, that would be in dangerous territory. We should treat it the same
* If we can't send a RREQ due to this limit being reached,
would we send a destination unreachable ICMP message to the
router client requesting the route?
as a congestive failure.
I am O.K. with this.
* My view of order of importance: RREP_Ack (to avoid being
blacklisted), RERR for undeliverable packet (to stop other
routers forwarding data to us), RREP, RREQ, RERR for newly
invalid routes.
I am O.K. with this also. It's a good idea, but no one ever asked
* How does the limit work? Wouldn't we need to check if we are
getting close to it, and then start discarding messages in
order of least importance as we get closer to the limit?
for it :-)
Regards,
Charlie P.
