[aodvv2-discuss] Re: Questions - Applicability Statement

  • From: Victoria Mercieca <vmercieca0@xxxxxxxxx>
  • To: "aodvv2-discuss@xxxxxxxxxxxxx" <aodvv2-discuss@xxxxxxxxxxxxx>
  • Date: Thu, 30 Apr 2015 14:54:34 +0100

Hi Charlie, hi all,

On Thu, Apr 16, 2015 at 5:37 PM, Charlie Perkins <
charles.perkins@xxxxxxxxxxxxx> wrote:

Hello Vicky,

Comments below...

On 4/16/2015 6:55 AM, Victoria Mercieca wrote:


This email contains questions about the applicability statement (Section
4 in Version 9a and Section 3 in Version 9b):

- 'If AODVv2 is the only protocol interacting with the route table' -
could someone expand on this paragraph as I didn't completely understand
it. The router will already send a request for any destination, if the
source of the packet to be sent is a router client. Does it mean it will
respond to a RREQ even if it is not responsible for the target?


This is a slightly complicated question. First, an AODVv2 router can also
be an OLSR router, an OSPF router, and/or a RIP router.
All of those other protocols are not prohibited by AODVv2, but by the same
token AODVv2 does not have (or need) mechanisms
to protect against incompatible updates. My opinion is that it's better
to NOT do this, but I don't want to get into the argument.
Second, an AODVv2 router can respond to any RREQ if it's obeying the
intermediate RREP specification, which I have rewritten
and am almost ready to resubmit.

I am afraid I didn't answer your question.


So the sentence currently says:
"When AODVv2 is the only protocol interacting with the forwarding table,
AODVv2 MAY be configured to perform route discovery for all unknown unicast
destinations. Such routers will reply for each address request."

Performing route discovery for all unknown unicast destinations - dont we
do that anyway? If a router client wishes to send a packet, does it matter
what the destination IP address is? We always send a RREQ.

The second sentence - replying for each address request - does this mean
the behaviour in the section on simple internet attachment?



- The bit about sparse traffic scenarios - is this because of unicast
routes? Does the statement still apply if we have subnets configured as
Router Clients?


No, it's because of the on-demand nature of the protocol. It is not
related to router clients.

Basically, "sparse traffic" ==> "fewer application launches" ==> fewer
routes needed ==> less control traffic.
Here, "sparseness" refers to traffic patterns, not volume of traffic.

My thinking was that if we have subnets configured as router clients, then
even though the RREQ and RREP contain unicast IP addresses, all routes
learned are to subnets, perhaps reducing the need for route requests in
future? I dont think this changes the statement that is already there
though.


- The paragraph about emergency scenarios - I cant help interpreting
this as "AODVv2 is great if you don't want any security" - seems a bit
strange to say this, especially with the changes we need to make to the
security section.


Strange but true. If your need for communication "HELP I AM ON FIRE HELP"
is stronger
than your need for security, then you will be very happy to have AODVv2
running.

However, obviously AODVv2 has to allow for strong security as well.

I've left this paragraph as is.

Regards,
Vicky.


Regards,
Charlie P.


Other related posts: