RFC 7767






Independent Submission                                     S. Vinapamula
Request for Comments: 7767                              Juniper Networks
Category: Informational                                     S. Sivakumar
ISSN: 2070-1721                                            Cisco Systems
                                                            M. Boucadair
                                                                  Orange
                                                                T. Reddy
                                                                   Cisco
                                                           February 2016


Application-Initiated Check-Pointing via the Port Control Protocol (PCP)

Abstract



   This document specifies a mechanism for a host to indicate via the
   Port Control Protocol (PCP) which connections should be protected
   against network failures.  These connections will then be subject to
   high-availability mechanisms enabled on the network side.

   This approach assumes that applications and/or users have more
   visibility about sensitive connections than any heuristic that can be
   enabled on the network side to guess which connections should be
   check-pointed.

Status of This Memo



   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This is a contribution to the RFC Series, independently of any other
   RFC stream.  The RFC Editor has chosen to publish this document at
   its discretion and makes no statement about its value for
   implementation or deployment.  Documents approved for publication by
   the RFC Editor are not a candidate for any level of Internet
   Standard; see Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc7767.











Vinapamula, et al.            Informational                     [Page 1]

RFC 7767                     HA through PCP                February 2016


Copyright Notice



   Copyright (c) 2016 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.

Table of Contents



   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
     1.1.  Note  . . . . . . . . . . . . . . . . . . . . . . . . . .   3
     1.2.  Requirements Language . . . . . . . . . . . . . . . . . .   3
   2.  Issues with the Existing Implementations  . . . . . . . . . .   4
   3.  CHECKPOINT_REQUIRED PCP Option  . . . . . . . . . . . . . . .   4
     3.1.  Format  . . . . . . . . . . . . . . . . . . . . . . . . .   4
     3.2.  Operation . . . . . . . . . . . . . . . . . . . . . . . .   5
   4.  Sample Use Cases  . . . . . . . . . . . . . . . . . . . . . .   7
   5.  Security Considerations . . . . . . . . . . . . . . . . . . .   8
   6.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .   9
   7.  References  . . . . . . . . . . . . . . . . . . . . . . . . .   9
     7.1.  Normative References  . . . . . . . . . . . . . . . . . .   9
     7.2.  Informative References  . . . . . . . . . . . . . . . . .   9
   Appendix A.  Additional Considerations  . . . . . . . . . . . . .  11
   Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . .  11
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  12

1.  Introduction



   The risk of Internet service disruption is critical in service
   providers and enterprise networking environments.  Such a risk is
   often mitigated with the introduction of active/backup systems.  Such
   designs not only contribute to minimize the risk of service
   disruption, but also facilitate maintenance operations (e.g., hitless
   hardware or software upgrades).

   In addition, the nature of some connections leads to the
   establishment and the maintenance of connection-specific states by
   some of the network functions invoked when the connection is
   established.  During active/backup failover in case of a network
   failure, the said states need to be check-pointed by the backup
   system.  Additional issues are discussed in Section 2.





Vinapamula, et al.            Informational                     [Page 2]

RFC 7767                     HA through PCP                February 2016


   Heuristics based on the protocol, mapping lifetime, etc., are used in
   the network to elect which connections need to be check-pointed
   (e.g., by means of high-availability (HA) techniques).  This document
   advocates for an application-initiated approach that would allow
   applications and/or users to signal to the network which of their
   connections are critical.

   Within this document, "check-pointing" refers to a process of state
   replication and synchronization between active and backup PCP-
   controlled devices.  When the active PCP-controlled device fails, the
   backup PCP-controlled device will take over all the existing
   established sessions that were check-pointed.  This process is
   transparent to internal hosts.

   This document specifies how PCP [RFC6887] can be extended to indicate
   which connection should be check-pointed for high availability
   (Section 3).  A set of use cases are provided for illustrative
   purposes in Section 4.  This document does not make any assumptions
   about the PCP-controlled device that will process the PCP-formatted
   signaling information from PCP clients.  These devices are likely to
   be flow aware.

   The approach in this document is aligned with the networking trends
   advocating for open network APIs to interact with applications/
   services (e.g., [RFC7149]).  For instance, the decision-making
   process about policy on the network side will be enriched with
   information provided by applications using PCP.

1.1.  Note



   The CHECKPOINT_REQUIRED PCP option (Section 3) is defined in the
   "Specification Required" range (see Section 6).  In order to be
   assigned a code point in that range, a permanent publication is
   required as per Section 4.1 of [RFC5226].  Publication of an RFC is
   an ideal means of achieving this requirement and also to ease
   interoperability.

   Note, this work was presented to the Port Control Protocol (PCP) WG,
   but there was no consensus to define this option in the "Standards
   Action" range despite positive feedback that was received from the
   working group.  Technical comments that were received during PCP
   meetings and those received on the mailing list were addressed.

1.2.  Requirements Language



   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC 2119 [RFC2119].



Vinapamula, et al.            Informational                     [Page 3]

RFC 7767                     HA through PCP                February 2016


2.  Issues with the Existing Implementations



   Regardless of the selected technology or design like HA-based
   designs, reliably securing connections is expensive in terms of
   memory, CPU usage, and other resources.  Also, check-pointing may not
   be required for all connections, as all connections may not be
   critical.  But, this leaves a challenge to identify what connections
   to check-point.

   Typically, this is addressed by identifying long-lived connections
   and check-pointing the state of only those connections that lived
   long enough, to the backup for service continuity.

   However, check-pointing long-lived connections raises the following
   issues:

   1.  It is hard for a network to identify (or guess) which connection
       is (business) critical.  This characterization is often customer-
       specific: a flow can be sensitive for a User #1, while it is not
       for another User #2.  Furthermore, this characterization can vary
       over time: a flow can be sensitive during hour X, while it is not
       during other times.

   2.  Heuristics are not deterministic.

   3.  A potentially long-lived connection may experience disruption
       upon failure of the active system, but before it is check-
       pointed.



   4.  A connection may not be long-lived but it may be critical, e.g.,
       for Voice over IP (VoIP) conversations.



   5.  Likewise, not all long-lived connections are deemed critical: for
       example, connections that pertain to free Internet services are
       usually considered not critical compared to the equivalent
       connections for paid services.  Only the latter need to be check-
       pointed.

3.  CHECKPOINT_REQUIRED PCP Option

3.1.  Format



   The solution is based on the assumption that an application or user
   is the best judge of which of its connections are critical.

   An application or user may explicitly identify the connections that
   need to be check-pointed by means of a PCP client, using the
   CHECKPOINT_REQUIRED option as described in Figure 1.



Vinapamula, et al.            Informational                     [Page 4]

RFC 7767                     HA through PCP                February 2016


   The entry to be backed up is indicated by the content of a MAP or
   PEER message.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |Option Code=192|  Reserved     |        Option Length          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

            Option Name: CHECKPOINT_REQUIRED
            Number: 192
            Purpose:  Indicate if an entry needs to be check-pointed.
            Valid for Opcodes: MAP, PEER
            Length: 0.
            May appear in: Request and response.
            Maximum occurrences: 1.

                 Figure 1: CHECKPOINT_REQUIRED PCP Option

   The description of the fields is as follows:

   o  Option Code: 192 (see Section 6).

   o  Reserved: This field is initialized as specified in Section 7.3 of
      [RFC6887].

   o  Option Length: 0.  This means no data is included in the option.

   An application or user can take advantage of this PCP option to
   explicitly indicate which of the connections need to be check-pointed
   and should not be disrupted.  The processing of this option by the
   PCP server will then yield the check-pointing of the corresponding
   states by the relevant devices or functions dynamically controlled by
   the PCP server.

   Communication between application/user and PCP client is
   implementation specific.

3.2.  Operation



   Support of the CHECKPOINT_REQUIRED option by PCP servers and PCP
   clients is optional.  This option (Code 192; see Figure 1) may be
   included in a PCP MAP or PEER request to indicate a connection is to
   be protected against network failures.

   There is a risk that every PCP client may wish to check-point every
   connection; this can potentially load the system.  Administration
   SHOULD restrict the number of connections that can be elected to be



Vinapamula, et al.            Informational                     [Page 5]

RFC 7767                     HA through PCP                February 2016


   backed up and the rate of check-pointing per network attachment point
   (e.g., Customer Premises Equipment (CPE), host).  To that aim, the
   PCP server should unambiguously identify the network attachment point
   a PCP client belongs to.  For example, the PCP server may rely on the
   PCP identity [RFC7652], the assigned prefix to a CPE or host, the
   subscriber-mask [PREFIX-BINDING], or other identification means.

   The PCP client includes a CHECKPOINT_REQUIRED option in a MAP or PEER
   request to signal that the corresponding mapping is to be protected.

   If the PCP client does not receive a CHECKPOINT_REQUIRED option in
   response to a PCP request that enclosed the CHECKPOINT_REQUIRED
   option, this means that either the PCP server does not support the
   option, or the PCP server is configured to ignore the option, or the
   PCP server cannot satisfy the request expressed in this option (e.g.,
   because of a lack of resources).

   If the CHECKPOINT_REQUIRED option is not included in the PCP client
   request, the PCP server MUST NOT include the CHECKPOINT_REQUIRED
   option in the associated response.

   When the PCP server receives a CHECKPOINT_REQUIRED option, the PCP
   server checks if it can honor this request depending on whether
   resources are available for check-pointing.  If there are no
   resources available for check-pointing, but there are resources
   available to honor the MAP or PEER request, a response is sent back
   to the PCP client without including the CHECKPOINT_REQUIRED option
   (i.e., the request is processed as any MAP or PEER request that does
   not convey a CHECKPOINT_REQUIRED option).  If check-pointing
   resources are still available and the quota for this PCP client has
   not been reached, the PCP server tags the corresponding entry as
   eligible to the HA mechanism and sends back the CHECKPOINT_REQUIRED
   option in the positive answer to the PCP client.

   To update the check-pointing behavior of a mapping maintained by the
   PCP server, the PCP client generates a PCP MAP or PEER renewal
   request that includes a CHECKPOINT_REQUIRED option to indicate this
   mapping has to be check-pointed or that doesn't include a
   CHECKPOINT_REQUIRED option to indicate this mapping does not need be
   check-pointed anymore.  Upon receipt of the PCP request, the PCP
   server proceeds with the same operations to validate a MAP or PEER
   request to update an existing mapping.  If validation checks are
   passed, the PCP server updates the check-point flag associated with
   that mapping accordingly (i.e., it is set if a CHECKPOINT_REQUIRED
   option was included in the update request or it is cleared if no
   CHECKPOINT_REQUIRED option was included), and the PCP server returns
   the response to the PCP client accordingly.




Vinapamula, et al.            Informational                     [Page 6]

RFC 7767                     HA through PCP                February 2016


   What information to check-point and how to check-point are outside
   the scope of this document and are left for implementations.  Also,
   the mechanism for users or applications to indicate check-pointing in
   a PCP request may be automatic, semiautomatic, or require human
   intervention.  This behavior is also left for application
   implementations.  For managed CPEs, a service provider may influence
   what connections are to be check-pointed.

   For honored requests, it is RECOMMENDED to check-point state on
   backup before a response is sent to the PCP client.

4.  Sample Use Cases

   Below are provided some examples for illustrative purposes:

   Example 1:  Consider a streaming service such as live TV
      broadcasting, or any other media streaming, that supports check-
      pointing signaling functionality.  Suppose this application is
      installed in three hosts A, B and C.  For A, the application is
      critical and should not be interrupted, while for B it is not.
      While for C, only some programs are of interest.  At the time of
      installing this application's software, corresponding preferences
      can be provisioned.  When the application starts streaming:

      *  All the flows associated with the streaming application are
         critical for A.  Limiting the number of flows to be backed up
         will ensure that host doesn't exceed the user's limit.

      *  For B, none of these flows are critical for check-pointing.
         The CHECKPOINT_REQUIRED option is not included in the PCP
         requests.

      *  For C, the user is invited to interact with the application by
         means of a configuration option that is provided to dynamically
         select which streaming to check-point, based on the user's
         interest.

   Example 2:  Consider a streaming service offered by a provider.
      Suppose three levels of subscriptions are offered by that
      provider, e.g., gold, silver, and bronze.  To guarantee a certain
      level of quality of service for each subscription, policies are
      configured such that:

      *  All flows associated with a gold subscription should be check-
         pointed.

      *  Only some flows associated with a silver subscription are
         check-pointed.



Vinapamula, et al.            Informational                     [Page 7]

RFC 7767                     HA through PCP                February 2016


      *  None of the flows associated with a bronze subscription are
         check-pointed.

      When a user invokes the streaming service, he/she may fall into
      one of those buckets, and according to the configured policy, his/
      her associated streaming flows are automatically check-pointed.
      Login credentials can be used as a trigger to determine the
      subscription level (and therefore the associated check-pointing
      behavior).

   Example 3:  Consider a VoIP application that is able to request that
      its flows be check-pointed.  No matter what is configured by the
      user, some calls such as emergency calls should be check-pointed.
      The application has to identify such calls.

   Example 4:  In the context of an enterprise network, applications are
      customized by the administrator.  Instructions about whether a
      CHECKPOINT_REQUIRED option is to be included are determined by the
      administrator.  Only the subset of applications identified by the
      administrator will make use of this option in conformance with the
      enterprise network's management policies.  Any misbehavior can be
      considered as abuse.

   In order to prevent every application from including a
   CHECKPOINT_REQUIRED option in its PCP requests, the following items
   are assumed:

   o  Applications may be delivered with some default settings for
      check-pointing, and these settings should be programmable by end
      user.

   o  Exposing and enforcing these settings is application specific.

   o  The end user may customize these settings based on the
      requirements.

5.  Security Considerations



   PCP-related security considerations are discussed in [RFC6887].

   The CHECKPOINT_REQUIRED option can be used by an attacker to identify
   critical flows; this is sensitive from a privacy standpoint.  Also,
   an attacker can cause critical flows to not be check-pointed by
   stripping the CHECKPOINT_REQUIRED option or by consuming the quota by
   adding the option to other flows.






Vinapamula, et al.            Informational                     [Page 8]

RFC 7767                     HA through PCP                February 2016


   These two issues can be mitigated if the network on which the PCP
   messages are to be sent is fully trusted.  Means to defend against
   attackers who can intercept packets between the PCP server and the
   PCP client should be enabled.  In some deployments, access control
   lists (ACLs) can be installed on the PCP client, PCP server, and the
   network between them, so those ACLs allow only communications between
   trusted PCP elements.  If the networking environment between the PCP
   client and the PCP server is not secure, PCP authentication [RFC7652]
   MUST be enabled.

   A network device can always override the end-user signaling, i.e.,
   what is signaled by the PCP client, if the instructions conflict with
   the network policies.

6.  IANA Considerations



   The following PCP Option Code has been allocated in the
   "Specification Required" range of the "PCP Options" registry
   (http://www.iana.org/assignments/pcp-parameters):

      192 CHECKPOINT_REQUIRED (see Section 3.1)



7.  References



7.1.  Normative References



   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <http://www.rfc-editor.org/info/rfc2119>.

   [RFC6887]  Wing, D., Ed., Cheshire, S., Boucadair, M., Penno, R., and
              P. Selkirk, "Port Control Protocol (PCP)", RFC 6887,
              DOI 10.17487/RFC6887, April 2013,
              <http://www.rfc-editor.org/info/rfc6887>.

   [RFC7652]  Cullen, M., Hartman, S., Zhang, D., and T. Reddy, "Port
              Control Protocol (PCP) Authentication Mechanism",
              RFC 7652, DOI 10.17487/RFC7652, September 2015,
              <http://www.rfc-editor.org/info/rfc7652>.

7.2.  Informative References



   [PREFIX-BINDING]
              Vinapamula, S. and M. Boucadair, "Recommendations for
              Prefix Binding in the Softwire DS-Lite Context", Work in
              Progress, draft-vinapamula-softwire-dslite-prefix-
              binding-12, October 2015.



Vinapamula, et al.            Informational                     [Page 9]

RFC 7767                     HA through PCP                February 2016


   [RFC5226]  Narten, T. and H. Alvestrand, "Guidelines for Writing an
              IANA Considerations Section in RFCs", BCP 26, RFC 5226,
              DOI 10.17487/RFC5226, May 2008,
              <http://www.rfc-editor.org/info/rfc5226>.

   [RFC7149]  Boucadair, M. and C. Jacquenet, "Software-Defined
              Networking: A Perspective from within a Service Provider
              Environment", RFC 7149, DOI 10.17487/RFC7149, March 2014,
              <http://www.rfc-editor.org/info/rfc7149>.










































Vinapamula, et al.            Informational                    [Page 10]

RFC 7767                     HA through PCP                February 2016


Appendix A.  Additional Considerations



   It was tempting to include additional fields in the option but this
   would lead to a more complex design that is not justified.  For
   example, we considered the following.

   o  Define a dedicated field to indicate a priority level.  This
      priority is intended to be used by the PCP server as a hint when
      processing a request with a CHECKPOINT_REQUIRED option.
      Nevertheless, an application may systematically choose to set the
      priority level to the highest value so that it increases its
      chance to be serviced!

   o  Return a more granular failure error code to the requesting PCP
      client.  However, this would require extra processing at both the
      PCP client and server sides for handling the various error codes
      without any guarantee that the PCP client would have its mappings
      check-pointed.

Acknowledgments

   Thanks to Reinaldo Penno, Stuart Cheshire, Dave Thaler, Prashanth
   Patil, and Christian Jacquenet for their comments.




























Vinapamula, et al.            Informational                    [Page 11]

RFC 7767                     HA through PCP                February 2016


Authors' Addresses



   Suresh Vinapamula
   Juniper Networks
   1194 North Mathilda Avenue
   Sunnyvale, CA  94089
   United States

   Phone: +1 408 936 5441
   Email: sureshk@juniper.net


   Senthil Sivakumar
   Cisco Systems
   7100-8 Kit Creek Road
   Research Triangle Park, NC  27760
   United States

   Phone: +1 919 392 5158
   Email: ssenthil@cisco.com


   Mohamed Boucadair
   Orange
   Rennes  35000
   France

   Email: mohamed.boucadair@orange.com


   Tirumaleswar Reddy
   Cisco Systems, Inc.
   Cessna Business Park, Varthur Hobli
   Sarjapur Marathalli Outer Ring Road
   Bangalore, Karnataka  560103
   India

   Email: tireddy@cisco.com













Vinapamula, et al.            Informational                    [Page 12]