Policies And how they affect IP Mobility

1
PoliciesAnd how they affect IP Mobility

• MOBOPTS RG, IETF 66
• Vijay Devarapalli, Jouni Korhonen,
• Rajeev Koodli, Telemaco Melia

2
IP Mobility in the IETF

• So far we have developed protocols that handle
  mobility securely
• Developed extensions and optimizations
• Fast handoffs
• Seamless handoffs
• Access control protocols
• But in deployment, there are other factors that
  influence a users mobility experience
• One of them is the policy the user is subject to
  when accessing a subscribed network

3
Policy

• What does a policy do?
• May govern what kind of service the user is
  allowed
• May govern which access network the user is
  allowed to attach
• E.g., network and realm selection
• May govern the type of mobility service the user
  is allowed
• May enforce a particular QoS for the user
  services
• May govern how long the user is allowed a
  particular service
• Typically stored centrally and delivered wherever
  needed based on the user connecting to the
  subscribed network
• Assumes a specific service model

4
Purpose

• To document the problems encountered in applying
  IP mobility due to policy constraints so that
• protocol designers can make better choices
• perhaps better policy architecture can evolve

5
Setting, terminology etc.

• Network where (some) control resides
• Mobile Node a typical MN, likely with with
  multiple physical interfaces, multiple
  subscriptions
• Multiple network access (e.g., WiFi and
  Cellular), multiple access providers, multiple
  service providers

6
Network - MN Tussle

• Network mainly interested in
• performance optimization (load balancing,vertical
  handovers)
• roaming across friendly boundaries (inter
  domain handovers)
• network controlled and network initiated
  handovers
• MN interested in
• Keeping the best connectivity (depending on radio
  conditions...)
• Make the best use of multiple interfaces
  (multihoming, multiple bindings...)
• Power saving
• How applications respond to mobility
• Unified model for both network side and terminal
  side feasible?

7
Some issues

• What does a MN need?
• Information about available networks
• What does the network want?
• Free dissemination? Controlled? Steered?
• How does MN regain connectivity?
• Slowly, especially across domains and access
  technologies
• What does the network want?
• Across domains, authentication enforced, some
  times multiple levels (access, service)

8
Some issues..

• What happens to MNs flows?
• Subject to policy
• What does the network want?
• Subject the flows to policy
• Whats the overlap of MNs wishes and Networks
  grants?
• Is the MN still in control?
• Can the MN still control its destiny?

9
Challenges

• If you are designing protocols
• Inconsistent policies across provider domains
• Performance
• Recognizing and acknowledging control points
• If you are designing policy
• Consistent policies that make sense across
  different access networks
• Deliver the policies to the right access node or
  a gateway in a timely manner
• A good mobility experience for the user