Internet DRAFT - draft-patil-mext-dmm-approaches
draft-patil-mext-dmm-approaches
Individual Submission B. Patil, Ed.
Internet-Draft Nokia
Intended status: Informational C. Williams
Expires: May 3, 2012 MCSR Labs
J. Korhonen
Nokia Siemens Networks
October 31, 2011
Approaches to Distributed mobility management using Mobile IPv6 and its
extensions
draft-patil-mext-dmm-approaches-02
Abstract
Mobility solutions at the IP layer have been specified in the IETF
for IPv4 and IPv6. These solutions include host and network based
mobility. All of the mobility protocols enable IP session continuity
by providing the mobile host with an IP address or prefix that
remains constant even as the host moves and attaches to different
access networks and points of attachment. Mobile hosts are anchored
at a gateway via a tunnel and the address/prefix provided to the host
via the gateway remains unchanged across mobility events. All IP
sessions initiated or terminated at a mobile host are anchored via
the gateway. A gateway centric approach raises certain concerns in
terms of cost and efficiency. A mobility model wherein the mobility
functions are distributed is a way of alleviating the concerns of a
gateway centric approach. This document considers ways to alleviate
anchored mobility issues with approaches that could be considered in
a deployment.
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on May 3, 2012.
Patil, et al. Expires May 3, 2012 [Page 1]
Internet-Draft DMM with MIP6 October 2011
Copyright Notice
Copyright (c) 2011 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Problem statement . . . . . . . . . . . . . . . . . . . . . . 4
4. Issues with current mobility models . . . . . . . . . . . . . 5
4.1. Backhauling all traffic to a centralized GW . . . . . . . 5
4.2. Latency Considerations . . . . . . . . . . . . . . . . . . 6
4.3. Inefficient Routing and signaling overhead . . . . . . . . 6
4.4. Scalability and cost . . . . . . . . . . . . . . . . . . . 6
5. Enhancements to improve mobility . . . . . . . . . . . . . . . 7
5.1. HMIPv6 . . . . . . . . . . . . . . . . . . . . . . . . . . 7
5.2. Dynamic assignment of HA . . . . . . . . . . . . . . . . . 7
5.3. Route Optimization . . . . . . . . . . . . . . . . . . . . 7
6. Distributed mobility - What does it imply . . . . . . . . . . 8
7. Approaches using current protocols for distributed mobility . 9
8. Potential future work . . . . . . . . . . . . . . . . . . . . 10
9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10
10. Security Considerations . . . . . . . . . . . . . . . . . . . 10
11. Summary and Conclusion . . . . . . . . . . . . . . . . . . . . 11
12. Informative References . . . . . . . . . . . . . . . . . . . . 11
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 12
Patil, et al. Expires May 3, 2012 [Page 2]
Internet-Draft DMM with MIP6 October 2011
1. Introduction
Mobility solutions at the IP layer have been specified in the IETF
for IPv4 and IPv6. These solutions include host and network based
mobility. All of the mobility protocols enable IP session continuity
by providing the mobile host with an IP address or prefix that
remains constant even as the host moves and attaches to different
access networks and points of attachment. Mobile hosts are anchored
at a gateway via a tunnel and the address/prefix provided to the host
via the gateway remains unchanged across mobility events. All IP
sessions initiated or terminated at a mobile host are anchored via
the gateway. There are issues and concerns with such a mobility
model which are discussed in this document. A mobility model wherein
the mobility functions are distributed is a way of alleviating the
concerns of a gateway centric approach. This document also considers
ways to alleviate anchored mobility issues with approaches that could
be considered in a deployment.
Mobile IPv6 as specified in [RFC6275] [RFC3776] is a host based
mobility protocol. It requires the MN to be anchored at a home
agent. The home agent assigns the MN an IPv6 address or prefix that
is static for the duration of the registration period. Similarly
Proxy Mobile IPv6 [RFC5213] is a network based mobility protocol in
which the mobility access gateway (MAG) assigns the MN a prefix
provided by the local mobility anchor (LMA) for the duration of a
valid registration. This prefix does not change across mobility
events. The home agent and LMA entities can be viewed as centralized
gateways. These gateways generally serve a large number of mobile
hosts. All traffic to/from mobile hosts associated with an HA/LMA is
routed through these gateways and as a result raises concerns such as
:
1. single point of failure,
2. backhauling traffic to the gateway,
3. latency as a result of backhauling and additional processing,
4. cost and complexity, etc.
These issues are discussed in further detail in the document. It
should also be noted that in addition to mobility for hosts, there is
also specifications that deal with networks that are mobile. Network
mobility is specified in [RFC3963]
The mobility working groups in the IETF have extended the basic
protocols to address various issues and concerns. Hierarchical
Mobile IP [RFC5380] and flow mobility [RFC6088], [RFC6089] are just a
Patil, et al. Expires May 3, 2012 [Page 3]
Internet-Draft DMM with MIP6 October 2011
few examples. Many of these extensions can be utilized in
deployments to alleviate the issues that arise from an anchored
mobility solution. A few approaches to how a distributed mobility
model could be deployed using current protocols and extensions are
also discussed in this document.
2. Terminology
Distributed Mobility
The term distributed mobility refers to an architecture in which
the mobility function is distributed across multiple levels in a
deployment. The mobility function could be provided by an access
point or base-station or it could be a part of the access network.
Distributed mobility would enable session continuity for hosts
while not requiring that they be anchored at a single gateway
(home agent) all the time.
3. Problem statement
The lack of support for mobility at the IP layer has been addressed
in IPv6 with the specification of Mobile IPv6 [RFC6275]. Various
extensions to the protocol such as support for multiple care-of-
addresses as well as the ability to operate while attached to an IPv4
network using Dual-stack Mobile IPv6 [RFC5555] have been specified.
The protocol has not been widely implemented or deployed as of date
for various reasons.
The Internet has evolved to support real-time applications such as
voice, multimedia streaming etc. These applications require low
latency as well as no (or minimal) interruptions when switching
interfaces or networks. Current IP mobility solutions based on
Mobile IPv6 are well suited for non-real-time applications which are
able to handle the delay which is caused by a mobile node doing a
handover between networks or switching interfaces. Optimizations to
support real time applications have also been specified such as
FMIPv6 [RFC5568]. The centralized gateway approach of Mobile IPv6
has multiple issues and raises concerns that are captured in this
document. One of the ideas is to move the mobility gateway closer to
the actual point of attachment. This has benefits in terms of
reduced latency but it also causes other issues such as the ability
to support mobility when the MN moves to a different access network
or the ability to do charging at a central node. Distributed
mobility is an approach that has some merit and worth studying
further. At the same time the issues that are driving the mobility
solutions towards a different model can be addressed by existing
Patil, et al. Expires May 3, 2012 [Page 4]
Internet-Draft DMM with MIP6 October 2011
protocols with various extensions. The problems of a centralized
gateway approach and reasons for considering distributed mobility
need to be deeply analyzed and understood before beginning work on
entirely new protocols for solving IP mobility.
4. Issues with current mobility models
Current mobility protocols have been designed with a stable
topologically correct anchoring gateway in mind. They just do not
tolerate mid-session anchor relocation. HMIP6, HA Switch, HA-
reliability and LMA Redirect are attempts in that direction but fail
or fall short.
In addition, one of the key deployment considerations of Mobile IPv6
is the location of each of the home agents or gateways, both
initially and over time. Each operator has unique requirements;
therefore, no single deployment model will suit all operators. The
operator's own organizational structure could also influence the
mobility architecture. Some operators have network OAM
responsibilities that are assigned geographically, while others use a
more centralized model. The deployment architecture that has been
traditionally put forth is to have centralized gateway elements where
all mobility control and data traffic is routed through them.
4.1. Backhauling all traffic to a centralized GW
A centralized home agent/gateway approach leads to backhauling all
traffic to the node which has unfavorable operational consequences.
The sheer volume of the aggregated throughput traffic to backhaul all
user data from a local aggregation anchor to centralized data centers
with home gateways can be expensive in many scenarios. With high
density deployments, the centralized architecture leads to heavy
backhaul utilization, and the inability to distribute load quickly
manifests unfavorably. In addition, local user traffic does not
remain local. User traffic must travel all the way to the
centralized gateway and back, even if the corresponding peer is
topologically closer.
In addition, a centralized gateway model increases the cost of
backhaul by preventing the off-loading of high-bandwidth services
locally. Instead high-bandwidth services have their traffic
backhauled to a centralized gateway in a data center. This will
increase the distances and possibly the capacity associated with any
backhaul.
Patil, et al. Expires May 3, 2012 [Page 5]
Internet-Draft DMM with MIP6 October 2011
4.2. Latency Considerations
While the support for Internet offload of user data can significantly
reduce the core network backhaul, the mobility management element may
be strategically positioned deeper in the network to efficiently
set-up and process the signaling and control including optional
policies. Such a hybrid architecture can provide for supporting a
mix of real-time and non-real-time broadband services. Real-time
applications can benefit from lower latencies by having data closer
to the subscriber and peers and not backhauled. Non-real-time
applications (such as e-mail) derive no such performance benefit and
may have a more centralized traffic approach.
Current mobility models handle offload cases poorly. A consideration
may be to clearly make a working toolbox for applications to select a
prefix with anchored mobility and a prefix without anchoring.
4.3. Inefficient Routing and signaling overhead
Inefficient routing mechanism of a completely centralized mobility
deployment approach causes QoS deterioration and may lead to heavy
network congestion in the core.
In the centralized approach only the HA and the CNs manage a nodes
mobility. Mobility signaling occurs each time a mobile node changes
its point-of-attachment regardless of the locality and amplitude of
its movement. As a consequence, the same level of signaling load is
introduced independently of the user's mobility pattern. For
example, if the HA and/or CNs are far from the MN, even if the MNs
movement is small, the mobility signaling messages travel across
several IP networks, the latencies of which reduce handover speed.
Furthermore, route optimization which supports direct routing from
CNs to the mobile node, generates excessive mobility messages and
adds a significant extra load to the network.
4.4. Scalability and cost
In a completely centralized Mobile IPv6-based deployment approach,
the home agent becomes a single point of failure. Also, a
distributed deployment approach may provide better overall capacity
and performance, but this must be weighed against the increase in
capital costs for deployment of local distributed gateways. In
addition, a completely centralized deployment model makes it
difficult to scale with a large number of mobile nodes. Scalability
costs are weighted from many perspectives such as the number of nodes
in the overall system, the geographic distance of the traffic, the
number of autonomous parties in the deployment approach and others.
Patil, et al. Expires May 3, 2012 [Page 6]
Internet-Draft DMM with MIP6 October 2011
5. Enhancements to improve mobility
Enhancements to the Mobile IPv6 protocol have been done to improve
mobile communications in certain scenarios so that mobility
operations are efficient and optimized.. A key area of enhancements
is in reducing the delays in the data path redirection operation that
is defined in Mobile IPv6 operations. Mobile IPv6 has adopted route
optimization and HMIPv6 to reduce the traversal of data traffic to
the mobile nodes new location changes in its point of attachment.
Delays in data traffic redirection will depend upon the location of
the anchor agent that performs the redirection. As such enhancements
focus on moving these anchor agents closer to the mobile node.
5.1. HMIPv6
Using Mobile IPv6, a mobile node sends location updates to any node
it corresponds with each time it changes its location, and at
intermittent intervals otherwise. This involves a lot of signaling
and processing, and requires a lot of resources. Furthermore,
although it is not necessary for external hosts to be updated when a
mobile nodes moves locally, these updates occur for both local and
global moves. Hierarchical Mobile IPv6 (HMIPv6)is designed to
enhance mobility support in MIPv6 and micro-mobility management. The
benefit of the HMIPv6 enhancement is to reduce the amount of
signaling required and to improve handoff speed.
The key concept behind HMIPv6 is to locally handle handovers by the
usage of an entity called the Mobility Anchor Point (MAP) located at
any level in a hierarchical network of routers. The major issue on
HMIPv6 is designing the MAP selection scheme that can reduce frequent
handover mobility signaling and improve handover performance.
5.2. Dynamic assignment of HA
Dynamic assignment of HA is an enhancement to reduce both the
signaling traffic and the data traffic to the home network. The
dynamic HA assignment may take into account the geographical
proximity of the HA to the mobile node. It may also consider
performance factors such as HA load-balancing or other criteria.
5.3. Route Optimization
Mobile IPv6 Route optimization is an enhancement to optimize the data
path between two communicating nodes despite changes in the IP
connectivity on the mobile node side. The data path reduction
between the communicating nodes helps to reduce one way packet delay
when both nodes are under the same localized domain and the mobility
gateway is far away. The process of reducing data path is referred
Patil, et al. Expires May 3, 2012 [Page 7]
Internet-Draft DMM with MIP6 October 2011
to as route optimization. Route optimization helps reduce the delay
and thus important for real-time applications. An enhanced version
of route optimization may also enable continued communications during
periods of temporary home-agent unavailability.
6. Distributed mobility - What does it imply
Mobility is a service that provides significant value to a network
operator. The ability to offer connectivity and services that work
seamlessly across mobility events such as the switching of an access
network type etc. creates a much superior end-user experience and
thereby a demand for such service. Cellular networks have offered
mobility for voice and messaging (short message service) since the
late 80s and early 90s. These networks have been evolving and are
now offering broadband data services and Internet connectivity. The
network architectures are also using Internet protocols and
technologies to a significant extent. Traditionally the
architectures of these networks has been hierarchical in nature.
While such an architecture served operators well in the past, it has
limitations when it comes to offering data services and Internet
connectivity. There is an effort to distribute functionality that
generally has resided in centralized gateways much more closer to the
edge of the network. The line between the access and core network is
fading and hence a need to rethink how mobility service is affected
in such an evolving architecture.
Distributed mobility is a way to deploy existing mobility solutions
that do not require a mobile host to be anchored at a gateway all the
time but instead be attached to different mobility agents/gateways in
the network depending on the access, location and other factors.
Session continuity via distributed mobility is expected to be on par
with that provided by an anchored mobility solution.
Does it require an entirely new approach to mobility architectures
that would be based on the goal of distributing mobility related
functions? It is an easy option to consider redesigning on a clean
sheet of paper. However this is not a pragmatic approach. It is
much more optimal to consider what are the issues that are created as
a result of a centralized gateway architecture and then develop
extensions to the protocols and, deployment models, that can address
those issues. The implications of distributed mobility architectures
on access and core networks needs to be also considered in any
design.
Patil, et al. Expires May 3, 2012 [Page 8]
Internet-Draft DMM with MIP6 October 2011
7. Approaches using current protocols for distributed mobility
We believe that most of the needed basic protocol functionality for
distributed mobility management is already there. What is missing
seem to be related to general system level design and lack of
mobility aware APIs for application developers. One of the simple
approaches for distributed mobility management is to avoid
traditional "anchored mobility" like Mobile IPv6 when possible and
rather use local (care-of) addresses for the communication. Use of
local addressing also implies less mobility related signaling load in
the network. For example [RFC5014] already provides means for an
application to explicitly request for a prefix that has mobility
characteristics (IPV6_PREFER_SRC_HOME) or a prefix that is local to
the current access network (IPV6_PREFER_SRC_COA). It is not
guaranteed that the IP stack in the MN would always respect the
suggestion received from the application. In general it is also
important that possible solutions in distributed mobility management
space requires minimal changes in mobile hosts.
Another aspect that is in interest of distributed mobility management
concentrates on allocating mobility anchors that are topologically
close to the MN. Existing protocols such as HMIPv6 [RFC5380] provide
a solution that is close what is needed. What might be needed in
addition is a mechanism to "chain" multiple MAP-domain to extend the
micro-mobility area, or provide another RFC5014 like prefix type
(IPV6_PREFER_SRC_MAP). We could also consider Mobile IPv6 + Proxy
Mobile IPv6 interactions Scenario A.1 in
[I-D.ietf-netlmm-mip-interactions] a similar solution. Finally, yet
another approach for exploiting locality are Proxy Mobile IPv6
localized routing solutions [I-D.ietf-netext-pmip6-lr-ps] which
allows bypassing the remote central Local Mobility Anchor when ever
possible and have a direct communication via closer to MNs Mobile
Access Gateways.
Home Agent Switch [RFC5142] extension to Mobile IPv6, Runtime LMA
assignment [I-D.ietf-netext-redirect] extension to Proxy Mobile IPv6
and Mobile IPv4 Dynamic HA Assignment [RFC4433] all provide solutions
to dynamically assign a mobility anchor to the MN. What is missing
from these solutions, is a protocol or rather a system level solution
for a "seamless mobility anchor relocation" during an existing
mobility session. However, that would be rather challenging due the
fact that a mobility anchor relocation usually implies topological
location chance in the network, which would also mean different
prefixes/subnetworks for home addresses from the IP routing point of
view. Within a reasonably small autonomous system or otherwise
restricted area maybe some kind of interior routing solution could be
used to assist mobility anchor relocation.
Patil, et al. Expires May 3, 2012 [Page 9]
Internet-Draft DMM with MIP6 October 2011
8. Potential future work
As the MEXT working group evolves and transitions to one that is
focused on dealing with distributed mobility, there is a need to
clearly understand the drivers for such an approach and whether these
could be dealt with via a framework that uses existing mobility
protocols and extensions and can be applied in a manner that deals
with those concerns.
One of the key efforts could be in understanding the key concerns
driving the need for a distributed mobility solution and identifying
various approaches using existing protocols and extensions to
overcome them.
1. Work on the generic solution for anchor relocation. This might
be a architecture describing work, rather than protocol work. I
believe we have most protocols already in place but not glued
together.
2. Work on address selection beyond RFC 5014 (with coloring i.e. the
end host stack knows properties of the prefix it got) and rapid
deprecation/renumbering of prefixes (needed when CoAs change and
applications try to use CoA for something local). This could
potentially be new protocol work an containers for coloring
prefixes (RA and DHCPv6) and how to handle local prefix
deprecation during handovers.
3. Work on localized mobility that does not involve signaling with
gateways or "mobility signaling". This could lead to work below
the IP layer, e.g. intra-AS mobility is handled using some
interior routing protocol enhancement.
9. IANA Considerations
This document has no requests to IANA.
10. Security Considerations
This document is a discussion of distributed mobility solutions.
Some of the approaches that are considered for deployment do have
security implications. However since the approaches being discussed
are based on existing mobility specifications developed within the
IETF, they have already been reviewed for security. This document
does not raise any new security concerns.
Patil, et al. Expires May 3, 2012 [Page 10]
Internet-Draft DMM with MIP6 October 2011
11. Summary and Conclusion
Distributed mobility is a way of deploying mobility protocols that
minimise the issues that arise from a centralized gateway centric
approach that comes from a hierarchical model. As the amount of
traffic in a network grows, operators are less willing to transport
all the traffic to a centralized gateway just for the sake of
enabling mobility. The mobility models have to evolve to meet the
changing environment of mobile networks and traffic patterns.
Using many of the extensions and protocols that have been defined for
Mobile IPv6 it is possible to deploy a mobility solution that meets
the criteria of distributed mobility architecture. The concerns fo a
centralized gateway approach can be addressed using deployment
techniques effectively.
12. Informative References
[I-D.ietf-netext-pmip6-lr-ps]
Liebsch, M., Jeong, S., and W. Wu, "PMIPv6 Localized
Routing Problem Statement",
draft-ietf-netext-pmip6-lr-ps-06 (work in progress),
March 2011.
[I-D.ietf-netext-redirect]
Korhonen, J., Gundavelli, S., Yokota, H., and X. Cui,
"Runtime LMA Assignment Support for Proxy Mobile IPv6",
draft-ietf-netext-redirect-12 (work in progress),
October 2011.
[I-D.ietf-netlmm-mip-interactions]
Giaretta, G., "Interactions between PMIPv6 and MIPv6:
scenarios and related issues",
draft-ietf-netlmm-mip-interactions-07 (work in progress),
October 2010.
[RFC3776] Arkko, J., Devarapalli, V., and F. Dupont, "Using IPsec to
Protect Mobile IPv6 Signaling Between Mobile Nodes and
Home Agents", RFC 3776, June 2004.
[RFC3963] Devarapalli, V., Wakikawa, R., Petrescu, A., and P.
Thubert, "Network Mobility (NEMO) Basic Support Protocol",
RFC 3963, January 2005.
[RFC4433] Kulkarni, M., Patel, A., and K. Leung, "Mobile IPv4
Dynamic Home Agent (HA) Assignment", RFC 4433, March 2006.
Patil, et al. Expires May 3, 2012 [Page 11]
Internet-Draft DMM with MIP6 October 2011
[RFC5014] Nordmark, E., Chakrabarti, S., and J. Laganier, "IPv6
Socket API for Source Address Selection", RFC 5014,
September 2007.
[RFC5142] Haley, B., Devarapalli, V., Deng, H., and J. Kempf,
"Mobility Header Home Agent Switch Message", RFC 5142,
January 2008.
[RFC5213] Gundavelli, S., Leung, K., Devarapalli, V., Chowdhury, K.,
and B. Patil, "Proxy Mobile IPv6", RFC 5213, August 2008.
[RFC5380] Soliman, H., Castelluccia, C., ElMalki, K., and L.
Bellier, "Hierarchical Mobile IPv6 (HMIPv6) Mobility
Management", RFC 5380, October 2008.
[RFC5555] Soliman, H., "Mobile IPv6 Support for Dual Stack Hosts and
Routers", RFC 5555, June 2009.
[RFC5568] Koodli, R., "Mobile IPv6 Fast Handovers", RFC 5568,
July 2009.
[RFC6088] Tsirtsis, G., Giarreta, G., Soliman, H., and N. Montavont,
"Traffic Selectors for Flow Bindings", RFC 6088,
January 2011.
[RFC6089] Tsirtsis, G., Soliman, H., Montavont, N., Giaretta, G.,
and K. Kuladinithi, "Flow Bindings in Mobile IPv6 and
Network Mobility (NEMO) Basic Support", RFC 6089,
January 2011.
[RFC6275] Perkins, C., Johnson, D., and J. Arkko, "Mobility Support
in IPv6", RFC 6275, July 2011.
Authors' Addresses
Basavaraj Patil (editor)
Nokia
6021 Connection drive
Irving, TX 75039
USA
Email: basavaraj.patil@nokia.com
Patil, et al. Expires May 3, 2012 [Page 12]
Internet-Draft DMM with MIP6 October 2011
Carl Williams
MCSR Labs
Palo Alto, CA 94306
USA
Email: carlw@mcsr-labs.org
Jouni Korhonen
Nokia Siemens Networks
Linnoitustie 6
FI-02600 Espoo
FINLAND
Email: jouni.nospam@gmail.com
Patil, et al. Expires May 3, 2012 [Page 13]
