Additional comments >>> below. What I'm proposing (spelled out a bit more) is
all win, and removes problems with the current design. I think this is a must
do.
--
Christopher Dearlove
Senior Principal Engineer
BAE Systems Applied Intelligence Laboratories
__________________________________________________________________________
T: +44 3300 467500 | E: chris(_dot_)dearlove(_at_)baesystems(_dot_)com
BAE Systems Applied Intelligence, Chelmsford Technology Park, Great Baddow,
Chelmsford, Essex CM2 8HN.
www.baesystems.com/ai
BAE Systems Applied Intelligence Limited
Registered in England & Wales No: 01337451
Registered Office: Surrey Research Park, Guildford, Surrey, GU2 7YP
-----Original Message-----
From: Jiazi Yi [mailto:ietf(_at_)jiaziyi(_dot_)com]
Sent: 09 May 2017 23:49
To: Dearlove, Christopher (UK)
Cc: ietf(_at_)ietf(_dot_)org; IETF-Announce; manet(_at_)ietf(_dot_)org;
draft-ietf-manet-olsrv2-multipath(_at_)ietf(_dot_)org;
manet-chairs(_at_)ietf(_dot_)org
Subject: Re: [manet] Last Call: <draft-ietf-manet-olsrv2-multipath-12.txt>
(Multi-path Extension for the Optimized Link State Routing Protocol version 2
(OLSRv2)) to Experimental RFC
----------------------! WARNING ! ---------------------- This message
originates from outside our organisation, either from an external partner or
from the internet.
Consider carefully whether you should click on any links, open any attachments
or reply.
Follow the 'Report Suspicious Emails' link on IT matters for instructions on
reporting suspicious email messages.
--------------------------------------------------------
Hi Chris,
Thanks a lot for the comments! Please check our reply inline:
On 9 May 2017, at 12:13, Dearlove, Christopher (UK)
<chris(_dot_)dearlove(_at_)baesystems(_dot_)com> wrote:
A few comments on this draft.
IPv6 specifies complete source routing. But this specification can only
consider what happens within the MANET. So for a packet from somewhere in the
MANET to somewhere well outside the MANET, the packet must be source routed
to the gateway between MANET and rest of Internet, and not source routed
after that. This I think should be explicitly mentioned. Whether that is
considered compliant with IPv6 I leave to others.
Good point. We added some text at the end of section 8.4" Datagram Processing
at the MP-OLSRv2 Originator"
7.1 SR_addr would better say "originator" address rather than
"network" address. (That makes it an address without a netmask, see
RFC 7181.)
fixed.
8.1 This is suggesting creating TC messages that have on neighbour addresses
but have only a SOURCE_ROUTE TLV. This is not the design I would have
suggested as consistent with how I would expect an extension to OLSRv2 to do
things. We need to consider two kinds of routers: those sending TC messages
anyway, those that (other than this extension) do not. In the former case you
could just add the SOURCE_ROUTE TLV to those TC messages it sends. Then that
information is maintained up to date. Routers that don't usually send TC
messages could send TC messages with just that TLV. But then there's an issue
over validity time. A parameter SR_HOLD_TIME_MULTIPLIER is introduced.
There's no need for that - you can simply incorporate that into the validity
time recorded in the message. That avoids a need to handle the two cases of
routers differently. There is then an oddity that you get some routers
sending TC messages with normal validity times and addresses, and some that
can be sen!
t less frequently with no addresses and longer validity times. But that's
suspect - note that it's not done in OLSRv2 for attached networks (another
reason to send TC messages although no neighbours need reporting). That's
because longer intervals make reacting to new routers joining (and network
reassembly after fragmentation) slow.
Rather a better design would simply be to add SOURCE_ROUTE TLV to normal TC
messages. When sending TC messages for just that reason, that could be just
the usual case, but you could allow as an option in this case to send less
frequently with validity time increased accordingly. When not using that
option, once a router needs to send a TC message, it could then decide to
report neighbours, increasing the topology distributed and allowing more
routes, this also being an option.
IIRC, we had a long discussion on this issue and produced the current text.
The purpose is to identify the routers that don’t send TC messages but support
source routing. To avoid unnecessary TC flooding, the interval is much longer
than the normal TC interval.
The normal TC messages (generated based on RFC7181) always have a SOURCE_ROUTE
TLV. But if we use the same validity time for both RFC7181 normal TC processing
and MP-OLSRv2 SR_ROUTE TLV processing, the valid time in the SR-OLSRv2 Router
Set would be much shorter than expected. A possible case is that, the router
stops sending normal TC messages, the corresponding entry in the SR-OLSRv2
Router set will soon expire.
Therefore, I think it’s reasonable to make use of the SR_HOLD_TIME_MULTIPLIER
to distinguish the valid time of normal TC message information and SR_ROUTE
information.
I don't agree, and I think what you are suggesting introduces a problem and
unnecessary overhead.
The problem is that everywhere in OLSRv2 we are careful to ensure that
parameters can be independently set, and that routers don't need to
coordinate to interoperate. (They may do better if they do, but that's a
refinement.) Here you are using an SR_HOLD_TIME_MULTIPLIER that the
receiver has to know is what the sender intends. And it's unnecessary. If
all that's in the TC message is the SOURCE_ROUTE TLV, you can just give
that a longer validity time, because the only thing that validity time will
impact on is the source routing status. If the router is also sending
normal TC messages and you send separate SOURCE_ROUTE TLV TC messages then
that would still be so. But that's inefficient, because if the router is
already sending TC messages, why send separate ones with added overhead,
when you can put the SOURCE_ROUTE TLV in the same TC message? That would
then (without SR_HOLD_TIME_MULTIPLIER, but that's a good thing because
SR_HOLD_TIME_MULTIPLIER is not good) give a shor!
ter validity time to the SOURCE_ROUTE TLV, but that's fine, because that
router will be sending more TC messages within that timescale anyway.
Furthermore, this also allows the router that's sending TC messages more
frequently to provide its information more responsively in the cases of nodes
joining and networks reassembling.
That gives two behaviours: routers sending TC messages anyway, just add a
SOURCE_ROUTE TLV, and routers not sending TC messages otherwise, just
include a SOURCE_ROUTE TLV and set the validity time according to what
schedule that router chooses to use - which can be at the same rate as the
other routers, or at a slower route, or (a new capability you don't have)
slowly, except if you learn of the existence of a new router in the network
you can send one or more responsive SOURCE_ROUTE TLV only TLVs to enable
that new router (or routers) to learn of the source routing quicker.
It's all win: combining messages, source set control, ability to do more
clever things if you want to. (You could probably even still send separate
TC messages with a SOURCE_ROUTE TLV when also sending normal TC messages if
you really wanted to, as an option I can't see wanting to use - although it
might introduce a minor problem that I haven't worked through the OLSRv2
specification to check, because it's unnecessary.)
The point is that what I suggest can gain everything you do, and allow
more, plus not breaking expected OLSRv2 behaviour.
8.3 second bullet. You should here (and possibly elsewhere) exclude routers
with routing willingness zero.
8.3 there seems to be an inconsistency. When operating proactively and no
multiple routes, drop the packet, but reactively use standard routing. The
latter seems more appropriate in the former case also.
9 CUTOFF_RATIO. Insists of strictly, but as defined earlier, may be >= 1`.
All fixed.
Thanks again for the valuable comments!
best
Jiazi
--
Christopher Dearlove
Senior Principal Engineer
BAE Systems Applied Intelligence Laboratories
______________________________________________________________________
____
T: +44 3300 467500 | E: chris(_dot_)dearlove(_at_)baesystems(_dot_)com
BAE Systems Applied Intelligence, Chelmsford Technology Park, Great Baddow,
Chelmsford, Essex CM2 8HN.
www.baesystems.com/ai
BAE Systems Applied Intelligence Limited Registered in England & Wales
No: 01337451 Registered Office: Surrey Research Park, Guildford,
Surrey, GU2 7YP
-----Original Message-----
From: manet [mailto:manet-bounces(_at_)ietf(_dot_)org] On Behalf Of The IESG
Sent: 20 April 2017 22:51
To: IETF-Announce
Cc: manet(_at_)ietf(_dot_)org;
draft-ietf-manet-olsrv2-multipath(_at_)ietf(_dot_)org;
manet-chairs(_at_)ietf(_dot_)org
Subject: [manet] Last Call: <draft-ietf-manet-olsrv2-multipath-12.txt>
(Multi-path Extension for the Optimized Link State Routing Protocol
version 2 (OLSRv2)) to Experimental RFC
----------------------! WARNING ! ---------------------- This message
originates from outside our organisation, either from an external partner or
from the internet.
Consider carefully whether you should click on any links, open any
attachments or reply.
Follow the 'Report Suspicious Emails' link on IT matters for instructions on
reporting suspicious email messages.
--------------------------------------------------------
*** WARNING ***
EXTERNAL EMAIL -- This message originates from outside our organization.
The IESG has received a request from the Mobile Ad-hoc Networks WG
(manet) to consider the following document:
- 'Multi-path Extension for the Optimized Link State Routing Protocol
version 2 (OLSRv2)'
<draft-ietf-manet-olsrv2-multipath-12.txt> as Experimental RFC
The IESG plans to make a decision in the next few weeks, and solicits final
comments on this action. Please send substantive comments to the
ietf(_at_)ietf(_dot_)org mailing lists by 2017-05-04. Exceptionally, comments
may be sent to iesg(_at_)ietf(_dot_)org instead. In either case, please
retain the beginning of the Subject line to allow automated sorting.
Abstract
This document specifies a multi-path extension for the Optimized Link
State Routing Protocol version 2 (OLSRv2) to discover multiple
disjoint paths, so as to improve reliability of the OLSRv2 protocol.
The interoperability with OLSRv2 is retained.
The file can be obtained via
https://datatracker.ietf.org/doc/draft-ietf-manet-olsrv2-multipath/
IESG discussion can be tracked via
https://datatracker.ietf.org/doc/draft-ietf-manet-olsrv2-multipath/bal
lot/
No IPR declarations have been submitted directly on this I-D.
_______________________________________________
manet mailing list
manet(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/manet
********************************************************************
This email and any attachments are confidential to the intended
recipient and may also be privileged. If you are not the intended
recipient please delete it from your system and notify the sender.
You should not copy it or use it for any purpose nor disclose or
distribute its contents to any other person.
********************************************************************