Internet Engineering Task Force                                  G. Chen
Internet-Draft                                              China Mobile
Intended status: Informational                               C. Williams
Expires: January 10, 2013                                     Consultant
                                                                 D. Wing
                                                          A. Yourtchenko
                                                     Cisco Systems, Inc.
                                                            July 9, 2012


            Happy Eyeballs Extension for Multiple Interfaces
               draft-ietf-mif-happy-eyeballs-extension-00

Abstract

   Currently the interface selection in multi-interface environment is
   exclusive - only one interface can be used at the time, frequently
   needing manual intervention.  Happy Eyeballs in MIF would make the
   selection process smoother by using the connectivity checks over a
   pre-filtered according to defined policy set of interfaces.  This
   would choose "best" interface with an automatic fallback.

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 January 10, 2013.

Copyright Notice

   Copyright (c) 2012 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



Chen, et al.            Expires January 10, 2013                [Page 1]


Internet-Draft             happy-eyeballs-mif                  July 2012


   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.  Problem Statement  . . . . . . . . . . . . . . . . . . . . . .  3
   3.  Happiness Parameters . . . . . . . . . . . . . . . . . . . . .  5
   4.  HE Behaviour in MIF  . . . . . . . . . . . . . . . . . . . . .  6
     4.1.  First Step, Filter . . . . . . . . . . . . . . . . . . . .  6
     4.2.  Second Step, Sort  . . . . . . . . . . . . . . . . . . . .  7
   5.  Implementation Framework . . . . . . . . . . . . . . . . . . .  8
   6.  Additional Considerations  . . . . . . . . . . . . . . . . . .  8
     6.1.  Usage Scope  . . . . . . . . . . . . . . . . . . . . . . .  8
     6.2.  Fallback Timeout . . . . . . . . . . . . . . . . . . . . .  8
     6.3.  Flow Continuity  . . . . . . . . . . . . . . . . . . . . .  9
   7.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . .  9
   8.  Security Considerations  . . . . . . . . . . . . . . . . . . .  9
   9.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . .  9
   10. References . . . . . . . . . . . . . . . . . . . . . . . . . .  9
     10.1. Normative References . . . . . . . . . . . . . . . . . . .  9
     10.2. Informative References . . . . . . . . . . . . . . . . . . 10
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 10
























Chen, et al.            Expires January 10, 2013                [Page 2]


Internet-Draft             happy-eyeballs-mif                  July 2012


1.  Introduction

   In multiple interface context, the problems raised by hosts with
   multiple interfaces have been discussed.  The MIF problem
   statement[RFC6418] described the various issues when using a wrong
   domain selection on a MIF node.  Happy Eyeballs (HE) [RFC6555]
   described how a dual-stack client can determine the functioning path
   to a dual-stack server.  It's using stateful algorithm to help
   applications quickly determine if IPv6 or IPv4 is the most optimal
   path to connect a server.  That is a good method to achieve smart
   path selection.  However, the assumption there is a single-homed
   context.  The interaction with multiple interfaces is deferred for
   further study.

   This memo has been proposed to extend happy eyeballs algorithm to fit
   into multiple interfaces context.  Several additional considerations
   have been elaborated to analyze the user demands and initiate HE-MIF
   connections.  This document describes a two-stage algorithm, which
   should allow a node with multiple interfaces both pick a satisfactory
   flow path, and identify the appropriate IP address family.


2.  Problem Statement

   The section enumerates several concrete use cases in existing
   networks.

   Case 1: WiFi is broken

   o  [Scenario] A MIF node has both 3G and WIFI interface.  When the
      node enters a WiFi area, a common practice would always prefer
      WiFi because it' cheap and fast-speed normally.

   o  [Problem] User assumes the wifi is working, because the node
      already got IP address from WiFi.  However, he can't run
      applications due to Internet connectivity being unavailable.  This
      could be an authentication required coming into play, or unstable
      Layer 2 conditions.  In order to figure out the problems, users
      have to go and turn off the WiFi manually.

   o  [Workaround] Users can indicate their desire with some setting on
      the phone.  For instance, they may prefer to wait a little bit of
      time but not forever.  After the timer is expired, users would
      finally give up the WiFi path and try to establish connection over
      3G path.  Users may won't want the wait time too short, because
      the 3G path for most people is more expensive than wifi path.

   Case 2: VPN (virtual private network) scenario



Chen, et al.            Expires January 10, 2013                [Page 3]


Internet-Draft             happy-eyeballs-mif                  July 2012


   o  [Scenario] In some cases, a node has multiple interface because of
      VPN.  Users would only have interests to connect a corporate
      network inside VPN.  While, connecting to Internet would work
      outside the VPN.

   o  [Problem] That is normally a implementation consideration that
      unmanaged interface may be considered less trustworthy than
      managed.  That results in trusted interfaces having the highest
      priority.  This setting may steer all traffic to VPN interface.
      When this is a traffic heading to a corporate site, everything is
      fine.  But sometimes, the connections out to Internet sites may
      suffer from long-distance path delays.

   o  [Workaround] It's desirable if routing could be bound to each
      interface.  However, a node following weak host model[RFC1122]
      takes routing tables as node-scoped.  One solution is to perform
      parallel IP connectivity checks before selecting an interface.

   Case 3: 3G/LTE tethering scenario

   o  [Scenario] Many mobile phones are equipped with software to offer
      tethered Internet access.  It shares their Internet connection
      with another Internet-capable mobile phone or other devices over
      Wi-Fi.  Apart from wifi link, there are some other links involved
      on the tethered phone, e.g. 3G access.

   o  [Problem] The WiFi link that tethered phone see is not free WiFi
      link, i.e. it might be 3G backhaul.  The policy of "always WiFi"
      leads to all traffic being sent over the tethering WiFi.  Usually,
      such tethering WiFi link puts sharing limitation to access nodes.
      It could cause contention on both that WiFi link and the backhaul
      3G link, while it be higher cost than going on the 3G that is
      built in the handset.

   o  [Workaround] To solve that, it is necessary for the node to be
      aware of not only the link layer information, but also services
      information, like billable or free.  That could help to facilitate
      the execution of the algorithm.  Same concern has been documented
      in Section 4.4 of [RFC6418])

   Case 4: Policy Conflict

   o  [Scenario] A node has WiFi and 3G access simultaneously.  In
      mobile network, IPv6-only may be preferable since IPv6 has the
      potential to be simpler than dual-stack.  WiFi access still remain
      on IPv4.





Chen, et al.            Expires January 10, 2013                [Page 4]


Internet-Draft             happy-eyeballs-mif                  July 2012


   o  [Problem] The problem is caused by policy confliction.  The
      transition to IPv6 is likely to encourage IPv6 and prefer
      IPv6[RFC3484].  If the 3G path has IPv6 on it and the WiFi does
      not, a suboptimal interface might be chosen from the cost saving
      perspective.

   o  [Workaround] Users interests should be well understood and
      considered before interface selection.  The different
      preconditions may impact subsequent behaviors.  Users concern
      about high-reliability or high-speed or less-cost should make
      different choice.  A flexible mechanism should be provided allow
      to make smart decision.


3.  Happiness Parameters

   To solve the problems, this section provide the design proposal for
   HE-MIF.  Two sets of "Happiness" parameters have been defined.  It
   serves upper applications and initiates HE-MIF connections to below
   level API subsequently.  Going through this two-stage process, MIF
   nodes could pick an appropriate interface which would correspond to
   user demands.  The two sets of "Happiness" parameters are called Hard
   set and Soft set respectively.

   o  Hard set: It contains parameters which have mandatory indications
      that interface behaviour should comply with.  This might provide
      an interface for applications constraints or delivering operator's
      policies.  Basically, parameters in Hard set should be easy-to-use
      and easy-to-understand.  The potential users would directly use
      those.

      *  Operator policies: operators would deliver the customized
         policies in particular network environments due to geo-location
         or services regulation considerations.  One example in 3GPP
         network is that operator could deliver policies from access
         network discovery and selection function (ANDSF).

      *  Application constraints: applications would express preferences
         which may not have a formally technical language , like "No 3G
         while roaming", "Only use free WiFi", etc.

   o  Soft set: It's a factor contributing to the best path.  The
      following is considered as for the justification.

      *  Next hop: [RFC4191] and DHCPv6 Route Option
         [I-D.ietf-mif-dhcpv6-route-option] allow configuration of
         specific routes to a destination.




Chen, et al.            Expires January 10, 2013                [Page 5]


Internet-Draft             happy-eyeballs-mif                  July 2012


      *  DNS selection: [I-D.ietf-mif-dns-server-selection] could
         configure nodes with information to indicate DNS server address
         for a particular namespace.

      *  Source address selection: the information provided by [RFC3484]
         would be considered.

      *  Other factors: There is a common practice may impact interface
         selection, e.g.  WiFi is preferable.  Such conventional
         experiences should also be considered.


4.  HE Behaviour in MIF

   Corresponding to the two sets of parameters, an HE-MIF node will take
   a two-step approach.  One is to do "hard" decision to synthesize
   policies from different actors (e.g., users and network operator).
   In a nutshell, that is filter which will exclude the interfaces from
   any further consideration.  The second is to adjust how we make a
   connection on multiple interfaces after the filter.  It's a sorting
   behaviour.  Those two things are described as following sub-sections.

4.1.  First Step, Filter

   One goal of filter is to reconcile multiple selection policies from
   users or operators.  Afterwards, the merged demands would be mapped
   to a set of candidate interfaces, which is judged as qualified.

   Decision on reconciliation of different policies will depend very
   much on the deployment scenario.  An implementation may not be able
   to determine priority for each policies without explicit
   configuration provided by users or administrator.  For example, an
   implementation may by default always prefer the WiFi due to cost
   saving consideration.  Whereas, users may dedicatedly prefer 3G
   interface to seek high-reliability or security benefits even to
   actively turn off WiFi interface.  The decision on mergence of
   policies may be made by implementation, by node administrators, even
   by other standards investigating customer behaviour.  However, it's
   worth to note that a demand from users may be normally considered
   higher priority than from other actors.

   The merged policies would serve as a filter principle doing iterate
   across the list of all known interfaces.  Qualified interface would
   be selected to sort processing at next step.







Chen, et al.            Expires January 10, 2013                [Page 6]


Internet-Draft             happy-eyeballs-mif                  July 2012


4.2.  Second Step, Sort

   Sort process would guarantee "best" interface selection with fallback
   capacities.  Two phases normally are involved in a whole session,
   i.e. name resolving and data session establishing.  Parameters in
   Soft set should considered at this stage.

   When the node initiates name requests, it should follow the
   instruction in [I-D.ietf-mif-dns-server-selection]if DNS server
   selection DHCP option is provided.  Otherwise, several alternative
   behaviour for DNS server selection described in Appendix A of
   [I-D.ietf-mif-dns-server-selection]maybe performed.

   Once a peer address was resolved, a connection would be intended to
   setup.  Heading to a destination, a particular interface may comply
   with soft parameter configuration, e.g. next hop[RFC4191]
   [I-D.ietf-mif-dhcpv6-route-option], source address selection[RFC3484]
   or common practice.  A particular interface should be treated with
   higher priority compared to others.  And, it should be choose to
   initiate the connection in advance.  This could avoid thrashing the
   network, by not (always) making simultaneous connection attempts on
   multiple interfaces.  After making a connection attempt on the
   preferred interface and failing to establish a connection within a
   certain time period (see Section 6.2), a HE-MIF implementation will
   decide to initiate connection attempt using rest of interfaces in
   parallel.  This fallback consideration may make subsequent connection
   attempts successful on non-preferable interface.

   The node would cache information regarding the outcome of each
   connection attempt.  Cache entries would be flushed periodically.  A
   system-defined timeout may take place to age the state.  Maximum on
   the order of 10 minutes defined in [RFC6555] is recommended to keep
   the interface state changes synchronizing with IP filmily states.  So
   long as new connections are being attempted by the MIF-node, such an
   implementation should occasionally make connection attempts using the
   soft-parameter's preferred interface, as it may have become
   functional again.

   If there are no specific soft-parameters provided, all interface
   should be equally treated.  The connections would initiate on several
   interface simultaneously.  The goal here is to provide fast
   connection for users, by quickly attempting to connect using one of
   interfaces.  Afterwards, the node would do the same caching and
   flushing process as described above.







Chen, et al.            Expires January 10, 2013                [Page 7]


Internet-Draft             happy-eyeballs-mif                  July 2012


5.  Implementation Framework

   The simplest way for the implementation is within the application
   itself.  The mechanism described in the document would not require
   any specific support from the operating system beyond the commonly
   available APIs that provide transport service.  It could also be
   implemented as high-level API approach, linking to MIF-API
   [I-D.ietf-mif-api-extension].  A number of enhancements could be
   added, making the use of the high-level APIs much more productive in
   building applications.


6.  Additional Considerations

6.1.  Usage Scope

   Connection-oriented transports (e.g., TCP, SCTP) could be directly
   applied as scoped in [RFC6555].  For connectionless transport
   protocols (e.g., UDP), it was also described "a similar mechanism can
   be used if the application has request/ response semantics (e.g., as
   done by Interactive Connectivity Establishment (ICE) to select a
   working IPv6 or IPv4 media path[RFC6157])."

6.2.  Fallback Timeout

   When the preferred interface was failed, HE-MIF would trigger
   fallback process to start connection initiation on several candidate
   interfaces.  It should set a reasonable wait time to comfort user
   experiences.  Aggressive timeouts may achieve quick interface
   handover, but at the cost of traffic that may be chargeable on
   certain networks.  E.g. the handover from WiFi to 3G would bring a
   bill to customers.  Considering the reasons, it is recommended to
   prioritize the input from users(e.g. real customers or applications)
   through UI.  For default-setting on a system, a hard error[RFC1122]
   in replied ICMP could serve as a trigger for the fallback process.
   When the ICMP soft error is present or non-response was received,
   it's recommended that the timeout should be large enough to allow
   connection retransmission.  [RFC1122] states that such timer MUST be
   at least 3 minutes to provide TCP retransmission.  Several minutes
   delay may not inappropriate for user experiences.  A widespread
   practice[RFC5461] sets 75 seconds to optimize connection process.

   [Editor note:It should be noted that 75 seconds is a worse case if
   host can't get UI input or hard error in ICMP.  More optimal timer
   may need to be defined depending on further investigation.]






Chen, et al.            Expires January 10, 2013                [Page 8]


Internet-Draft             happy-eyeballs-mif                  July 2012


6.3.  Flow Continuity

   Interface changing should only happen at the beginning of new session
   in order to keep flow continuity for ongoing TCP session.  Dynamic
   movement of traffic flows are beyond the scope of this document.


7.  IANA Considerations

   This memo includes no request to IANA.


8.  Security Considerations

   The security consideration is following the statement in [RFC6555]and
   [RFC6418].


9.  Acknowledgements

   The authors would like to thank Margaret Wasserman, Hui Deng, Erik
   Kline, Stuart Cheshire and Teemu Savolainen for their helpful
   comments.


10.  References

10.1.  Normative References

   [I-D.ietf-mif-dns-server-selection]
              Savolainen, T., Kato, J., and T. Lemon, "Improved
              Recursive DNS Server Selection for Multi-Interfaced
              Nodes", draft-ietf-mif-dns-server-selection-10 (work in
              progress), June 2012.

   [RFC1122]  Braden, R., "Requirements for Internet Hosts -
              Communication Layers", STD 3, RFC 1122, October 1989.

   [RFC3484]  Draves, R., "Default Address Selection for Internet
              Protocol version 6 (IPv6)", RFC 3484, February 2003.

   [RFC4191]  Draves, R. and D. Thaler, "Default Router Preferences and
              More-Specific Routes", RFC 4191, November 2005.

   [RFC6555]  Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with
              Dual-Stack Hosts", RFC 6555, April 2012.





Chen, et al.            Expires January 10, 2013                [Page 9]


Internet-Draft             happy-eyeballs-mif                  July 2012


10.2.  Informative References

   [I-D.ietf-mif-api-extension]
              Liu, D., Lemon, T., Ismailov, Y., and Z. Cao, "MIF API
              consideration", draft-ietf-mif-api-extension-00 (work in
              progress), March 2012.

   [I-D.ietf-mif-dhcpv6-route-option]
              Dec, W., Mrugalski, T., Sun, T., and B. Sarikaya, "DHCPv6
              Route Options", draft-ietf-mif-dhcpv6-route-option-04
              (work in progress), February 2012.

   [I-D.ietf-mif-problem-statement]
              Blanchet, M. and P. Seite, "Multiple Interfaces and
              Provisioning Domains Problem Statement",
              draft-ietf-mif-problem-statement-15 (work in progress),
              May 2011.

   [RFC5461]  Gont, F., "TCP's Reaction to Soft Errors", RFC 5461,
              February 2009.

   [RFC6157]  Camarillo, G., El Malki, K., and V. Gurbani, "IPv6
              Transition in the Session Initiation Protocol (SIP)",
              RFC 6157, April 2011.

   [RFC6418]  Blanchet, M. and P. Seite, "Multiple Interfaces and
              Provisioning Domains Problem Statement", RFC 6418,
              November 2011.


Authors' Addresses

   Gang Chen
   China Mobile
   53A,Xibianmennei Ave.,
   Xuanwu District,
   Beijing  100053
   China

   Email: phdgang@gmail.com











Chen, et al.            Expires January 10, 2013               [Page 10]


Internet-Draft             happy-eyeballs-mif                  July 2012


   Carl Williams
   Consultant
   El Camino Real
   Palo Alto, CA  94306
   USA

   Email: carlw@mcsr-labs.org


   Dan Wing
   Cisco Systems, Inc.
   170 West Tasman Drive
   San Jose, CA  95134
   USA

   Email: dwing@cisco.com


   Andrew Yourtchenko
   Cisco Systems, Inc.
   De Kleetlaan, 7
   Diegem  B-1831
   Belgium

   Email: ayourtch@cisco.com


























Chen, et al.            Expires January 10, 2013               [Page 11]