Transcript PPT Version

Virtual network interface for
multiple interfaces in a mobile
node using Mobile IPv6
draft-hong-multipleif-mn-pb-statement-00.txt
draft-hong-virtualif-mn-mipv6-00.txt
Y-G Hong, J-C Lee, J-S Park, H-J Kim
ETRI
IETF-64 monami6 Meeting @Vancouver
1
Goals of these drafts
• Identify some problems of multiple interfaces in a
MN or MR
• Mobile IPv6-specific issues
• General network issues
• Consideration for Heterogeneous Environments
• Propose a virtual network interface model for
multiple interfaces in a MN or MR
2
Problem Statements
• Mobile IPv6 Specific Issues
– Determination of network movement
– Receiving RA message from different interfaces
• General Network Issues
– Updating the relation between a destination address and
a network interface
• Consideration for Heterogeneous Environments
– Using Mobile IPv6 and NEMO in IPv4 based network
3
Issues in monami6 ML for Problem
Statements (1/3)
• For Mobile IPv6 Specific Issues
– About MIPL, the current code has some parts related to interface
preference mechanisms (10/25, by Romain)
– The problem statement sticks too much to implementation issues, not
to generic issues related to multihoming. (10/25, by Romain)
– The problem you describe in your draft is an implementation specific
issue, as pointed out by Romain Kuntz. In draft-montavont, we have a
section "considerations for MIP6 implementation". So, at best some
text could be added there. (10/26, by Thierry)
– Using RA reception to determine a movement or even to determine the
availibility/unavailibility of an interface is an information that we want to
consider in general. Note that in our revision, we changed media
detection to failure detection, in order to consider all failure that can
occur on the entire path between the MN and its CN. It can be on the
media (interface connected or not at L2, or no IPv6 prefix on the link),
or a failure on the path between the MN and the HA, or between the
HA and the CN. So, for me, RA reception is part of failure detection.
We may have to add text to explicitly mention that it is considered.
Anyway, I think we have lot of work to clarify failure detection.(10/28,
by Nicolas)
4
Issues in monami6 ML for Problem
Statements (2/3)
• For General Network Issues
– The problem yoiu describe in your draft is related, when considering a
mobile node, to the HoA address, and you ranged it into a "generic
issue" because it can apply to a fixed node with multiple interfaces,
right? I think this is a problem in which SHIM6 is qualified (how to
change the locator, but not the identifier). I agree that our text in draftmontavont should be improved in a fashion similar to draft-ietf-nemomultihoming-issues where such issues are well described. (10/26 by
Thierry)
– Even though this problem that end node should update the relation
between destination address and specific network interface to prevent
dissonance between active network interface and changed IP address
was happened in fixed end node, I do not agree that the SHIM6 should
qualify this problem. AFAIK, SHIM6 is considering environment where
an end node has multiple paths (multiple IPv6 address prefix) by
connecting to different ISP that means site multihoming. So site
multihomed an end node does not have necessarily multiple interfaces
and it only has to obtain multiple addresses from different site exit
routers. But the mobile node equipped multiple interfaces are down for
discussion in MONAMI6 WG clearly. Therefore, if we have consensus,
which this problem must be solved for using multiple interfaces, I think
5
that the MONAMI6 is more suitable place than SHIM6. (10/27 By
Taewan You)
Issues in monami6 ML for Problem
Statements (3/3)
• For Heterogeneous Environments Issues
– About Heterogeneous Environments, you speak about v4-only access
on CDMA interface. I believe you never roam on the CDMA interface,
so a tunneling system would fit to your problem (l2tp, or DTCP). If not
you may take a look at draft-ietf-mip6-nemo-v4traversal-00.txt (10/25
by Romain)
– The concern described in your draft is valid and is addressed in draftietf-mip6-nemo-v4traversal-00.txt. Note that I do not consider it has a
multihoming issues at all: it is right that nodes equipped with multiple
interfaces may be more concerned with this issue, but this could also
happen for a node with a single interface (for instance, a node with a
WI-FI interface roaming between an IPv6 wireless LAN and an IPv4
wireless LAN. We already came across this problem (10/26 By Thierry)
6
Issues in monami6 ML for Problem
Statements
• For general Issues
– Actually, I kind of agree that more text could be useful in draftmontavont., in the same fashion as in draft-ietf-nemo-multihomingissues under the issue "Media Detection". However, I think we disagree
on the classification as I wouldn't consider it as a MIP6-specific
issue.(10/26 by Thierry)
7
Architecture of virtual I/F model
8
Operations of virtual I/F model
• When a Mobile IPv6/NEMO starts
– Virtual interface is configured to send/receive packets
– Path between one real network interface and virtual
interface is made
• When sending packets
– Virtual interface -> Real network interface
• When receiving packets
– Real network interface -> Virtual interface
9
Issues in monami6 ML for Virtual
Interface model
•
•
•
The virtual interface model is a good idea when you plan to use
several interfaces, but not at the same time. What if you want to
use several interfaces at the same time? (10/25, by Romain)
To use multiple interfaces simultaneously, you have to register the
CoA of each interface to the Home Agent(s) (ie send several BUs),
handle policies for each interface etc. You may also have events
on several interfaces at the same time. How can you handle this
with the virtual interface model? (10/27, by Romain)
I was speaking about several "physical interface" that you want to
use _simultaneously_, ie for example by registering the CoA of
_each_ interface (as proposed in draft-wakikawa-mobileipmultiplecoa for example). I am not sure how you can deal with that
with only one virtual interface. (10/28, by Romain)
10
Next Steps ?
• Our Problem Statements can be added into some
related documents and we can improve Nicolas’s
document
• Virtual Interface model can be one solution to handle
multiple interface node without modifying Mobile
IPv6 and NEMO
• If there is some consensus, we would like to improve
the virtual interface document
• Hopefully part of this WG
11