Internet Engineering Task Force (IETF)                          N. Akiya
Request for Comments: 7419                               M. Binderberger
Updates: 5880                                              Cisco Systems
Category: Informational                                        G. Mirsky
ISSN: 2070-1721                                                 Ericsson
                                                          December 2014


    Common Interval Support in Bidirectional Forwarding Detection

Abstract

  Bidirectional Forwarding Detection (BFD) requires that messages be
  transmitted at regular intervals and provides a way to negotiate the
  interval used by BFD peers.  Some BFD implementations may be
  restricted to only support several interval values.  When such BFD
  implementations speak to each other, there is a possibility of two
  sides not being able to find a common value for the interval to run
  BFD sessions.

  This document updates RFC 5880 by defining a small set of interval
  values for BFD that we call "Common Intervals" and recommends
  implementations to support the defined intervals.  This solves the
  problem of finding an interval value that both BFD speakers can
  support while allowing a simplified implementation as seen for
  hardware-based BFD.  It does not restrict an implementation from
  supporting more intervals in addition to the Common Intervals.

Status of This Memo

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

  This document is a product of the Internet Engineering Task Force
  (IETF).  It represents the consensus of the IETF community.  It has
  received public review and has been approved for publication by the
  Internet Engineering Steering Group (IESG).  Not all documents
  approved by the IESG are 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/rfc7419.








Akiya, et al.                 Informational                     [Page 1]

RFC 7419             Common Interval Support in BFD        December 2014


Copyright Notice

  Copyright (c) 2014 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  . . . . . . . . . . . . . . . . . . . . . . . .   2
  2.  The Problem with Few Supported Intervals  . . . . . . . . . .   3
  3.  Well-Defined, Common Intervals  . . . . . . . . . . . . . . .   4
  4.  Security Considerations . . . . . . . . . . . . . . . . . . .   4
  5.  References  . . . . . . . . . . . . . . . . . . . . . . . . .   5
    5.1.  Normative References  . . . . . . . . . . . . . . . . . .   5
    5.2.  Informative References  . . . . . . . . . . . . . . . . .   5
  Appendix A.  Why Some Values Are in the Common Interval Set . . .   6
  Appendix B.  Timer Adjustment with Non-identical Interval Sets  .   6
  Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . .   8
  Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .   8

1.  Introduction

  The Bidirectional Forwarding Detection (BFD) standard [RFC5880]
  describes how to calculate the transmission interval and the
  detection time.  However, it does not make any statement about how to
  solve a situation where one BFD speaker cannot support the calculated
  value.  In practice, this may not have been a problem as long as
  software-implemented timers were used and as long as the granularity
  of such timers was small compared to the interval values being
  supported, i.e. as long as the error in the timer interval was small
  compared to 25 percent jitter.

  In the meantime, requests exist for very fast interval values, down
  to 3.3 msec for the MPLS Transport Profile (MPLS-TP).  At the same
  time, the requested scale for the number of BFD sessions is
  increasing.  Both requirements have driven vendors to use Network
  Processors (NP), Field Programmable Gate Arrays (FPGAs), or other
  hardware-based solutions to offload the periodic packet transmission
  and the timeout detection in the receive direction.  A potential



Akiya, et al.                 Informational                     [Page 2]

RFC 7419             Common Interval Support in BFD        December 2014


  problem with this hardware-based BFD is the granularity of the
  interval timers.  Depending on the implementation, only a few
  intervals may be supported, which can cause interoperability
  problems.  This document proposes a set of interval values that
  should be supported by all implementations.  Details are laid out in
  the following sections.

2.  The Problem with Few Supported Intervals

  Let's assume vendor "A" supports 10 msec, 100 msec, and 1 sec
  interval timers in hardware, and vendor "B" supports every value from
  20 msec onward, with a granularity of 1 msec.  For a BFD session, "A"
  tries to set up the session with 10 msec while "B" uses 20 msec as
  the value for RequiredMinRxInterval and DesiredMinTxInterval.  Rx and
  Tx are negotiated as described in [RFC5880], which is 20 msec in this
  case.  However, system "A" is not able to support the 20 msec
  interval timer.  Multiple ways exist to resolve the dilemma, but none
  of them is without problems.

  a.  Realizing that it cannot support 20 msec, system "A" sends out a
      new BFD packet advertising the next larger interval of 100 msec
      with RequiredMinRxInterval and DesiredMinTxInterval.  The new
      negotiated interval between "A" and "B" is then 100 msec, which
      is supported by both systems.  However, the problem is that we
      moved from the 10/20 msec range to 100 msec, which has far
      deviated from operator expectations.

  b.  System "A" could violate [RFC5880] and use the 10 msec interval
      for the Tx direction.  In the receive direction, it could use an
      adjusted multiplier value M' = 2 * M to match the correct
      detection time.  Now, in addition to the fact that we explicitly
      violate [RFC5880], there may be the problem that system "B" drops
      up to 50% of the packets; this could be the case when "B" uses an
      ingress rate policer to protect itself and the policer would be
      programmed with an expectation of 20 msec receive intervals.

  The example above could be worse when we assume that system "B" can
  only support a few timer values itself.  Let's assume "B" supports 20
  msec, 300 msec, and 1 sec.  If both systems would adjust their
  advertised intervals, then the adjustment ends at 1 sec.  The example
  above could even be worse when we assume that system "B" can only
  support 50 msec, 500 msec, and 2 sec.  Even if both systems walk
  through all of their supported intervals, the two systems will never
  be able to agree on an interval to run any BFD sessions.







Akiya, et al.                 Informational                     [Page 3]

RFC 7419             Common Interval Support in BFD        December 2014


3.  Well-Defined, Common Intervals

  The problem can be reduced by defining interval values that are
  supported by all implementations.  Then, the adjustment mechanism
  could find a commonly supported interval without deviating too much
  from the original request.

  In technical terms, the requirement is as follows: a BFD
  implementation should support all values in the set of Common
  Interval values that are equal to or larger than the fastest (i.e.,
  lowest) interval the particular BFD implementation supports.

  This document defines the set of Common Interval values to be: 3.3
  msec, 10 msec, 20 msec, 50 msec, 100 msec, and 1 sec.

  In addition, both a 10 sec interval and multiplier values up to 255
  are recommended to support graceful restart.

  The adjustment is always towards larger (i.e., slower) interval
  values when the initial interval proposed by the peer is not
  supported.

  This document is not adding new requirements with respect to the
  precision with which a timer value must be implemented.  Supporting
  an interval value means advertising this value in the
  DesiredMinTxInterval and/or RequiredMinRxInterval field of the BFD
  packets and providing timers that are reasonably close.  [RFC5880]
  defines safety margins for the timers by defining a jitter range.

  How is the Common Interval set used exactly?  In the example above,
  vendor "A" has a fastest interval of 10 msec and thus would be
  required to support all intervals in the Common Interval set that are
  equal or larger than 10 msec, i.e., it would support 10 msec, 20
  msec, 50 msec, 100 msec, and 1 sec.  Vendor "B" has a fastest
  interval of 20 msec and thus would need to support 20 msec, 50 msec,
  100 msec, and 1 sec.  As long as this requirement is met for the
  common set of values, then both vendor "A" and "B" are free to
  support additional values outside of the Common Interval set.

4.  Security Considerations

  This document does not introduce any additional security concerns.
  The security considerations described in the BFD documents, [RFC5880]
  and others, apply to devices implementing the BFD protocol,
  regardless of whether or not the Common Interval set is implemented.






Akiya, et al.                 Informational                     [Page 4]

RFC 7419             Common Interval Support in BFD        December 2014


5.  References

5.1.  Normative References

  [RFC5880]  Katz, D. and D. Ward, "Bidirectional Forwarding Detection
             (BFD)", RFC 5880, June 2010,
             <http://www.rfc-editor.org/info/rfc5880>.

5.2.  Informative References

  [G.8013_Y.1731]
             International Telecommunications Union, "OAM functions and
             mechanisms for Ethernet based networks", ITU-T
             Recommendation G.8013/Y.1731, November 2013.

  [GR-253-CORE]
             Telcordia Technologies, Inc., "Synchronous Optical Network
             (SONET) Transport Systems: Common Generic Criteria",
             GR-253-CORE Issue 05, October 2009.
































Akiya, et al.                 Informational                     [Page 5]

RFC 7419             Common Interval Support in BFD        December 2014


Appendix A.  Why Some Values Are in the Common Interval Set

  The list of Common Interval values is trying to balance various
  objectives.  The list should not contain too many values, as more
  timers may increase the implementation costs.  On the other hand,
  fewer values produces larger gaps and adjustment jumps.  More values
  in the lower interval range are thus seen as critical to support
  customer needs for fast detection in setups with multiple vendors.

  o  3.3 msec: required by MPLS-TP, to support the defect detection
     time of 10 msec from [GR-253-CORE].

  o  10 msec: general consensus is to support 10 msec.  Multiple
     vendors plan to or do already implement 10 msec.

  o  20 msec: basically avoids a larger gap in this critical interval
     region.  Still allows 50-60 msec detect and restore (with
     multiplier of 2) and covers existing software-based
     implementations.

  o  50 msec: widely deployed interval.  Supporting this value reflects
     the reality of many BFD implementations today.

  o  100 msec: similar to 10 msec, this value allows the reuse of
     [G.8013_Y.1731] implementations, especially hardware.  It supports
     a large number of 100 msec sessions with multiplier 9 (9 x 100
     msec), which could be replacing of 3 x 300 msec configurations
     used by customers to have a detection time slightly below 1 sec
     for VoIP setups.

  o  1 sec: as mentioned in [RFC5880].  While the interval for Down
     packets can be 1 sec or larger, this document recommends use of
     exactly 1 sec to avoid interoperability issues.

  The recommended value for large intervals is 10 sec, allowing for a
  timeout of 42.5 minutes with a multiplier of 255.  This value is kept
  outside the Common Interval set, as it is not required for normal BFD
  operations that occur in the sub-second range.  Instead, the expected
  usage is for graceful restart, if needed.

Appendix B.  Timer Adjustment with Non-identical Interval Sets

  [RFC5880] implicitly assumes that a BFD implementation can support
  any timer value equal to or above the advertised value.  When a BFD
  speaker starts a Poll Sequence, then the peer must reply with the
  Final (F) bit set and adjust the transmit and detection timers
  accordingly.  With contiguous software-based timers, this is a valid
  assumption.  Even in the case of a small number of supported interval



Akiya, et al.                 Informational                     [Page 6]

RFC 7419             Common Interval Support in BFD        December 2014


  values, this assumption holds when both BFD speakers support exactly
  the same interval values.

  But what happens when both speakers support intervals that are not
  supported by the peer?  An example is router "A" supporting the
  Common Interval set plus 200 msec, while router "B" supports the
  Common Intervals plus 300 msec.  Assume both routers are configured
  and run at 50 msec.  Now, router A is configured for 200 msec.  We
  know the result must be that both BFD speakers use 1 sec timers, but
  how do they reach this endpoint?

  First, router A sends a packet with 200 msec.  The P bit is set
  according to [RFC5880].  The Tx timer stays at 50 msec, the detection
  timer is 3 * 200 msec:

     (A) DesiredTx: 200 msec, MinimumRx: 200 msec, P-bit
     Tx: 50 msec, Detect: 3 * 200 msec

  Router B now must reply with an F bit.  The problem is B is
  confirming timer values that it cannot support.  The only setting to
  avoid a session flap would be

     (B) DesiredTx: 300 msec, MinimumRx: 300 msec, F-bit
     Tx: 50 msec, Detect: 3 * 300 msec

  immediately followed by a P-bit packet, as the advertised timer
  values have been changed:

     (B) DesiredTx: 300 msec, MinimumRx: 300 msec, P-bit
     Tx: 50 msec, Detect: 3 * 300 msec

  This is not exactly what Section 6.8.7 of [RFC5880] states about the
  transmission rate.  On the other hand, as we will see, this state
  does not last for long.  Router A would adjust its timers based on
  the received Final bit:

     (A) Tx: 200 msec, Detect: 3 * 1 sec

  Router A is not supporting the proposed 300 msec and would use 1 sec
  instead for the detection time.  It would then respond to the
  received Poll Sequence from router B using 1 sec, as router A does
  not support the Max(200 msec, 300 msec):

     (A) DesiredTx: 1 sec, MinimumRx: 1 sec, F-bit
     Tx: 200 msec, Detect: 3 * 1 sec






Akiya, et al.                 Informational                     [Page 7]

RFC 7419             Common Interval Support in BFD        December 2014


  followed by its own Poll Sequence, as the advertised timer values
  have been changed:

     (A) DesiredTx: 1 sec, MinimumRx: 1 sec, P-bit
     Tx: 200 msec, Detect: 3 * 1 sec

  Router B would adjust its timers based on the received Final bit

     (B) Tx: 300 msec , Detect: 3 * 1 sec

  and would then reply to the Poll Sequence from router A:

     (B) DesiredTx: 300 msec, MinimumRx: 300 msec, F-bit
     Tx: 1 sec, Detect: 3 * 1 sec

  which finally makes router A adjust its timers:

     (A) Tx: 1 sec, Detect: 3 * 1 sec

  In other words, router A and B go through multiple Poll Sequences
  until they reach a commonly supported interval value.  Reaching such
  a value is guaranteed by this document.

Acknowledgments

  We would like to thank Sylvain Masse and Anca Zamfir for bringing up
  the discussion about the Poll Sequence, and Jeffrey Haas for helping
  find the fine line between "exact" and "pedantic".

Authors' Addresses

  Nobo Akiya
  Cisco Systems

  EMail: [email protected]


  Marc Binderberger
  Cisco Systems

  EMail: [email protected]


  Greg Mirsky
  Ericsson

  EMail: [email protected]




Akiya, et al.                 Informational                     [Page 8]