Internet Engineering Task Force                                   X. Wei
INTERNET-DRAFT                                                     L.Zhu
Intended Status: Standards Track                     Huawei Technologies
Expires: April 11, 2015                                           L.Deng
                                                            China Mobile
                                                         October 8, 2014

                       Tunnel Congestion Feedback


   This document describes a mechanism to calculate congestion of a
   tunnel segment based on RFC 6040 recommendations, and a feedback
   protocol by which to send the measured congestion of the tunnel from
   egress to ingress router. A basic  model for measuring tunnel
   congestion and feedback is described, and a protocol for carrying the
   feedback data is outlined.

Status of this Memo

   This Internet-Draft is submitted to IETF in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF), its areas, and its working groups.  Note that
   other groups may also distribute working documents as

   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."

   The list of current Internet-Drafts can be accessed at

   The list of Internet-Draft Shadow Directories can be accessed at

Copyright and License Notice

   Copyright (c) 2014 IETF Trust and the persons identified as the
   document authors. All rights reserved.

Wei                      Expires April 11, 2015                 [Page 1]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   ( 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. Conventions and Terminology . . . . . . . . . . . . . . . . . .  4
     2.1 Conventions  . . . . . . . . . . . . . . . . . . . . . . . .  4
     2.2  Terminology . . . . . . . . . . . . . . . . . . . . . . . .  4
   3. Problem Statement . . . . . . . . . . . . . . . . . . . . . . .  5
     3.1 3GPP network scenario  . . . . . . . . . . . . . . . . . . .  6
     3.2 Network Function Virtualization Scenario . . . . . . . . . .  7
     3.3 Data Center Tenancy Scenario . . . . . . . . . . . . . . . .  9
   4. Congestion Control Model  . . . . . . . . . . . . . . . . . . .  9
     4.1 Congestion Calculation . . . . . . . . . . . . . . . . . . . 10
     4.2 Data Information . . . . . . . . . . . . . . . . . . . . . . 12
     4.3 Congestion Feedback  . . . . . . . . . . . . . . . . . . . . 12
     4.4 Congestion Control . . . . . . . . . . . . . . . . . . . . . 13
   5. Congestion Feedback Protocol  . . . . . . . . . . . . . . . . . 13
     5.1 Properties of Candidate Protocol . . . . . . . . . . . . . . 13
     5.2 IPFIX Extensions for Congestion Feedback . . . . . . . . . . 14
     5.3 Other Protocols  . . . . . . . . . . . . . . . . . . . . . . 18
   6. Benefits  . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
   7. Security Considerations . . . . . . . . . . . . . . . . . . . . 18
   8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . . 18
   9. References  . . . . . . . . . . . . . . . . . . . . . . . . . . 19
     9.1  Normative References  . . . . . . . . . . . . . . . . . . . 19
     9.2  Informative References  . . . . . . . . . . . . . . . . . . 19
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 20

Wei                      Expires April 11, 2015                 [Page 2]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

1. Introduction

   In current practice of Internet protocol, encapsulation of IP headers
   is always the technical proposal for overlay networking scenarios.
   For example, mobile network are designed to encapsulate inner IP
   header and application layer header chain through IP header, UDP
   header and GTP-U header. It is also designed to fulfill the mobility,
   QoS control, bearer management and other specific application of the
   mobile network. Some organization's private network encrypt IP header
   by Internet tunnel solutions with private key or certification
   approaches to setup VPN (virtual private network) over WAN (wide area

   Congestion is the situation that traffic input exceeds throughput of
   any segment of transmission path, which can result from
   transportation constraints and interface/processor overload. In
   general, congestion seen as the cause of packet loss or   unexpected
   delay to network end points. End to end congestion protocols (e.g.
   ECN [RFC 3168] and ECN handling for tunneling   scenario [RFC6040])
   are discussed in IETF.

   In IP header encapsulation cases, IP headers should be carried over
   transportation protocol like TCP or UDP, which influents the explicit
   congestion control feedback, since the receiver should mark ECN in
   TCP acknowledgment. On the other hand, packet loss and performance
   degradation should not be recognized by network elements, for
   instance the tunnel ingress and egress entity, when network segment
   is encapsulated by IP header and UDP header chain. That causes
   management problem when tunnel segment is considered as an
   independent administration domain, and network operator intents to
   keep network operation reliable.

   This document describes a mechanism for feedback of congestion
   observed in IP tunnels usages. Common tunnel deployments such as
   mobile backhaul networks, VPNs and other IP-in-IP tunnels can be
   congested as a result of sustained high load.

   Network providers use a number of methods to deal with high load
   conditions including proper network dimensioning, policies for
   preferential flow treatment and selective offloading among others.
   The mechanism proposed in this document is expected to complement
   them and provide congestion information that to allow making better,
   policies and decisions.

   The model and general solution proposed in chapter 4 consist of
   identifying congestion marks set in the tunnel segment, and feeding
   back the congestion information from the egress to the ingress of the
   tunnel. Measuring congestion of a tunnel segment is based on counting

Wei                      Expires April 11, 2015                 [Page 3]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   outer packet CE marks for packets that have ECT marks in the inner
   packet. This proposal depends on statistical marking of congestion
   and uses the method described in RFC 6040 [RFC6040], Appendix C.

   In chapter 5 the desired properties of the congestion information
   conveying protocol are outlined, and IPFIX [RFC5101] as a candidate
   protocol for these extensions is explored further.

2. Conventions and Terminology

2.1 Conventions

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in RFC 2119 [RFC2119]

2.2  Terminology

   Tunnel:        A channel over which encapsulated packets traverse
                  across a network.

   Encapsulation: The process of adding control information when it
                  passes through the layered model.

   Encapsulator:  The tunnel endpoint function that adds an outer IP
                  header to tunnel a packet, the encapsulator is
                  considered as the "ingress" of the tunnel.

   Decapsulator:  The tunnel endpoint function that removes an outer IP
                  header from a tunneled packet, the decapsulator is
                  considered as the "egress" of the tunnel.

   Outer header:  The header added to encapsulate a tunneled packet.

   Inner header:  The header encapsulated by the outer header.

   E2E:           End to End.

   VPN:           Virtual Private Network is a technology for using the
                  Internet or another intermediate network to connect
                  computers to isolated remote computer networks that
                  would otherwise be inaccessible.

   GRE:           Generic Routing Encapsulation.

   IPFIX              IP Flow Information Export. An IETF protocol to export
                  flow information from routers and other devices.

Wei                      Expires April 11, 2015                 [Page 4]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   RED        Random Early Detection

   NFV          Network Functions Virtualization is an alternative design
                  approach for building complex IT applications,
                  particularly in the telecommunications and service
                  provider industries, that virtualizes entire classes
                  of function into building blocks that may be
                  connected, or chained, together to create services.

   VNF          Virtualized Network Function may consist of one or more virtual
                  machines running different software and processes,
                  which form the building blocks for NFV.

   SFC          Service Function Chain is a group of connected VNF in a specific
                  sequence/map using NFV approach, in order to deliver a
                  specific service.

3. Problem Statement

   Network traffic congestion control plays a significant role in
   network performance management, and sustaining congestion  could
   impact subscriber's experience. Currently the solution of network
   congestion problem mainly focuses on end-to-end method, i.e. ECN
   [RFC3168], and the traffic sender are in charge of reducing traffic
   rates in case of network congested. But sometimes it's not always
   reliable to dependent on end hosts to solve the congestion situation,
   because some end hosts may not support ECN, or even ECN is supported
   by end hosts some traffics, e.g. UDP-based traffic, may not support

   Though the congestion happens in operator's network, in case that the
   congestion information is transparent to operator, network
   administration would be hard to take action to control the network
   traffic of reason to network congestion. To improve the performance
   of the network, it's better for operator to take network congestion
   situation into network traffic management.

   Many kinds of tunnels are widely deployed in current networks, even
   in some scenarios all traffics transmitted through designated

   Because the ingress and egress of tunnel are usually deployed by
   operator, so it's easy for operator to execute operator's policy, for
   example gating, flow control and dropping. The tunnel feedback
   mechanism should be feasible for operator to collect network
   congestion information in encapsulation segment. After obtaining

Wei                      Expires April 11, 2015                 [Page 5]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   congestion information, operator could make policy at tunnel ingress
   for traffic management taking these information into consideration.

   ECN handling mechanisms in RFC 6040 specifies how ECN should be
   handled for tunneling. In addition, RFC 6040, Appendix C provides
   guidance to calculate congestion experienced in the tunnel itself.
   However, there is no standardized mechanism by which the congestion
   information inside the tunnel can be fed back from egress to ingress

   In the following sub-sections, some network tunnel scenarios are

3.1 3GPP network scenario

   Tunnels, including GRE [RFC2784], GTP [TS29.060], IP-in-IP [RFC2003]
   or IPSec [RFC4301] etc, are widely deployed in 3GPP networks. And in
   3GPP network tunnels are used to carry end user flows within the
   backhaul network such as shown in Figure 1.

   IP backhaul networks such as those of mobile networks are provisioned
   and managed to provide the subscribed levels of end user service.
   These networks are traffic engineered, and have defined mechanisms
   for providing differentiated services and QoS per user or flow.
   Policy to configure per user flow attributes in these networks have
   traditionally been based on monitoring and static configuration.

   Currently, these networks are increasingly used for applications that
   demand high bandwidth. The nature of the flows and length of end user
   sessions can lead to significant variability in aggregate bandwidth
   demands and latency. In such cases, it would be useful to have a more
   dynamic feedback of congestion information. In addition, eNB, SGW and
   PGW are administrated by one mobile operator, mobile backhaul to
   carry IP/UDP/GTP encapsulation is regally administrated by back haul
   service operator. This aggregate congestion feedback could be used to
   determine flow handling and admission control.
                  +-|---+           +------+         +------+
       +--+       |     |  Tunnel1  |      | Tunnel2 |      |  Ext
       |UE|-(RAN)-| eNB |===========| S-GW |=========| P-GW |--------
       +--+       |     |    RAN    |      |  Core   |      |Network
                  +-+---+  Backhaul +---+--+ Network +---+--+

        Figure 1: Example - Mobile Network and Tunnels

Wei                      Expires April 11, 2015                 [Page 6]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

3.2 Network Function Virtualization Scenario

   Telecoms networks contain an increasing variety of proprietary
   hardware appliances, leading to increasing difficulty in lauching new
   network services, as well as the complexity of integrating and
   deploying these appliances in a network.

   Network Functions Virtualisation (NFV) aims to address these problems
   by decoupling the software from dedicated hardware platforms to a
   range of industry standard server hardware for various network
   services, through IT virtualization technology that can be moved to,
   or instantiated in, various locations in the network as required. In
   this way, it is expected to provide significant benefits for network
   operators (reduced expenditures for network construction and
   maintenance) and their customers (shortened time-to-market for new
   network services).

   Furthermore, service functions are preferred to be deployed and
   managed in a data center manner, rather than being inserted on the
   data-forwarding path between communicating peers as today. SFC WG is
   currently working on a new framework to cope with this highly dynamic
   routing problem for a network service, which requires that the
   relevant data traffic be traversing a group of virtualized network
   function nodes (VNFs), each of which could be applied at any layer
   within the network protocol stack (network layer, transport layer,
   application layer, etc.). [SFC]

   As shown in Figure 2, in a SFC-enabled domain (e.g. with or across
   network operator's deployed data centers), a PDP (Policy Decision
   Point) is the central entity which is responsible for maintaining SFC
   Policy Tables (rules for the boundary nodes on deciding which IP flow
   to traverse which service function path), and enforcing appropriate
   policies in SF Nodes and SFC Boundary Nodes. Beginning at the Ingress
   node, at each hop of a given service function path (as decided by a
   matched SFC policy rule/map), if the next function node is not an
   immediate (L3) neighbor, packet are encapsulated and forwarded to
   correspondent downstream function node, as shown in Figure 3.

Wei                      Expires April 11, 2015                 [Page 7]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

                   . . . . . . . . . . . . . . . . . . . . . . . . .
                   . SFC Policy Enforcement                        .
                   .             +-------+                         .
                   .             |       |-----------------+       .
                   .     +-------|  PDP  |                 |       .
                   .     |       |       |-------+         |       .
                   .     |       +-------+       |         |       .
                   . . . | . . . . . | . . . . . | . . . . | . . . .
                   . . . | . . . . . | . . . . . | . . . . | . . . .
                   .     |           |           |         |       .
                   .     v           v           v         v       .
                   . +---------+ +---------+ +-------+ +-------+   .
                   . |SFC_BN_1 | |SFC_BN_n | | SF_1  | | SF_m  |   .
                   . +---------+ +---------+ +-------+ +-------+   .
                   . SFC-enabled Domain                            .
                   . . . . . . . . . . . . . . . . . . . . . . . . .

        Figure 2: SFC Policy Enforcement Scheme

                                 Network Service
           +----------+           +----------+           +----------+
           |   VNF#1  | tunnel#1  |   VNF#2  | tunnels   |   VNF#n  |
           | Instance |-----------| Instance |- ... ... -| Instance |
           +----------+           +----------+           +----------+
                                       | Virtualization
           |                Virtualization Platform                 |

        Figure 3: Example - Mobile Network service chaining and Tunnels

   However, using VNFs running commodity platforms can introduce
   additional points of failure beyond those inherent in a single
   specialized server, and therefore poses additional challenges on
   reliability. [VNFPOOL] proposes using pooling techniques in response,
   which requires maintaining a backup mapping among running VNF
   instances for a given service function, and choosing from them for a
   specific data flow. It is clear that it would be helpful to make more
   efficient use of network capacity in case of local congestion, if the
   choice is based on the ECN feedback as well as the running status
   and/or physical resources accommodation of a candidate VNF instance.

Wei                      Expires April 11, 2015                 [Page 8]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

3.3 Data Center Tenancy Scenario

   In the scenario of data center of multi-tenant, network resource
   would be shared between more than one tenants, and in order to
   provide functional isolation and at the same time guarantee
   scalability for tenants, the tunnel based isolation mechanisms, e.g.
   VxLAN and STT etc, are provided.

   In the scenario described above, hypervisor or vSwitch would act as
   tunnel endpoint for the traffic between VMs, and tunnels are agnostic
   to VMs, in other words, the congestion indication information such as
   ECN flag marked by network entity of data center are agnostic to VMs.
   To deal with this situation, two solutions could be used:

   Solution 1: Using tunnel translation, hypervisor or vSwitch marks the
   inner IP header according to ECN flag in outer IP header before
   transmits packets to VM.

   Solution 2: Using the congestion control mechanism provided in this
   document between hypervisors or vSwitchs to do congestion control for
   VMs' traffic.

4. Congestion Control Model

   In this section, the basic congestion control model will be provided,
   and each detailed aspect of this model will also be introduced in the
   following subsection.

   The congestion control model provides network administrator with a
   method to manage the data traffic in its network domain. The basic
   model consists of the following components: Ingress, Egress,
   Feedback, Meter, Collector and Manager.

   As shown in Figure 4, network traffic enters the tunnel through
   tunnel ingress, passing through en-route routers, which will mark
   packets according to ECN mechanism as specified in RFC3168, to tunnel
   egress; the egress collects the congestion level information
   encountered in tunnel and feeds back it to the corresponding ingress;
   after receiving congestion information, the ingress takes actions to
   control the traffic that passing through the path between the ingress
   and egress to reduce the congestion level in the tunnel.

   At egress, a module named Meter is used to estimate the congestion
   level in the tunnel as described in the section above. A congestion
   information feedback module, called Feedback, is used to control the
   congestion information feedback procedure.

   The metering module named Meter in the Egress node accounts the

Wei                      Expires April 11, 2015                 [Page 9]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   congestion marks it receives. The Feedback module calculates the
   amount of congestion and feeds back the congestion information to the
   Ingress node. The Collector at the Ingress receives the congestion
   information which is fed back from the Feedback module. The Manager
   implements functions such as admission control and traffic
   engineering according to the congestion level experienced in tunnel
   to control the traffic to reduce the congestion level, the detailed
   actions taken by the Manager are out of the scope of current

                            congestion feedback signal
             +-----#-------+                        +------#----+
             |     #       |                        |      #    |
             |     #       |                        |      #    |
             |     V       |                        |      #    |
             | +---------+ |   +--------------+     | +--------+|
             | |Collector| |   |              |     | |Meter   ||
      traffic| +---------+ |   |              |     | +-----+--+|traffic
      ======>| |Manager  | |======================> | |Feedback||======>
             | +---------+ |   |   Routers    |     | +--------+|
             |             |   | (ECN-enabled)|     |           |
             +-------------+   +--------------+     +-----------+

        Figure 4: Basic Feedback Model

   To support traffic management and congestion information feedback in
   tunnel, there are mainly three issues that this document discusses:
   calculation of congestion level information, feeding back the
   congestion information from egress to ingress, and implementation of
   congestion control. The tunnel ingress/egress are assumed to be
   compliant with RFC6040 and the tunnel interior routers are compliant
   with RFC3168.

   In addition, it should be noted that these tunnels may carry ECT or
   Not-ECT traffic. A well defined mechanism for aggregate congestion
   calculation should be able to work in the presence of all kinds of
   traffic and would benefit from a common feedback mechanism and

4.1 Congestion Calculation

   This section discusses how to calculate congestion level experienced
   in the tunnel, an example of how to calculate congestion level is
   provided. In this document calculation of congestion in the tunnel is
   based on the method described in RFC 6040, Appendix C.

Wei                      Expires April 11, 2015                [Page 10]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   The egress can calculate congestion using moving averages. The
   proportion of packets not marked in the inner header that have a CE
   marking in the outer header is considered to have experienced
   congestion in the tunnel. Note that the packets are ECN capable and
   not congestion-marked before tunnel. Since routers implementing RED
   randomly select a percentage of packets to mark, this method can be
   effectively used to expose congestion in the tunnel.

   When the ingress is  RFC6040 compliant, the packets collected by
   egress can be divided into to 4 categories, shown in figure 5. The
   tag before "|" stands for ECN field in outer  header; and the tag
   after "|" stands for ECN field in inner header.

   "Not-ECN|Not-ECN" indicates traffic that does not support ECN, for
   example UDP and Not-ECT marked TCP; "CE|CE" indicates  ECN capable
   packets that have CE-mark before entering the tunnel; "CE|ECT"
   indicates ECN capable packets that are CE-marked in the tunnel;
   "ECT|ECT" indicates ECN capable packets that have not experienced
   congested in tunnel (or outside the tunnel).

      |     Not-ECN|Not-ECN      |
      |          CE|CE           |
      |          CE|ECT          |
      |         ECT|ECT          |

        Figure 5: ECN marking categories by outer/inner packet

   Out of the total number of packets, if the quantity of CE|ECT packets
   is A, the quantity of ECT|ECT packets is B, then the congestion level
   (C) can be calculated as follows:


   As an example, consider 100 packets to calculate the moving average
   as shown in RFC 6040, Appendix C. Say that there are 12 packets that
   have CE|ECT marks indicating that they have experienced congestion in
   the tunnel. And, there are 58 packets that have ECT|ECT marks
   indicating that there was no congestion in either the tunnel or
   elsewhere. The egress can calculate congestions as:

Wei                      Expires April 11, 2015                [Page 11]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

                        C = 12/ (12 + 58)
                          = 12/70 (17% congestion)

4.2 Data Information

   This section discusses congestion-related information that should be
   conveyed from egress to ingress.

   (1)Congestion volume. The information indicating the how much
   congestion has been experienced in the tunnel by traffic passing
   through the tunnel. Because there are both ECT packets and Not-ECT
   packets passing through the tunnel network, and in case of
   congestion, the ECT packets would be CE-marked instead of dropped and
   tunnel egress can be aware of these CE-marked packets; but Not-ECT
   packets would be dropped and tunnel egress cannot be aware of these
   dropped packets, so it's hard for egress to calculate the precise
   number of congested packets. According to the analysis in subclause
   4.1, the congestion volume is preferred in the form of percentage,
   e.g. 17.14%.

   (2)Egress identifier. To control the traffic congestion in certain
   tunnel, the ingress needs to have the knowledge of which traffic
   should be controlled, especially for the case that the ingress
   establishes tunnels with different egresses. So the egress identifier
   should be transmitted together with congestion volume to ingress.
   This identifier is usually the identifier of the tunnel or the
   address of tunnel egress.

4.3 Congestion Feedback

   This sub-section focuses on the discussion of feedback procedure. The
   congestion feedback procedure conveys congestion status from egress
   to ingress. The discussion of feedback protocol will be discussed in
   the next section.

   To reduce the overload, caused by this procedure, on network
   especially in case the feedback signal goes through the same path as
   data traffic, the feedback will only occur when congestion happens.
   In other words, egress doesn't send feedback signal if there is no
   congestion happens. Also egress will ignore ephemeral congestion and
   only feed back congestion information if the congestion level goes
   higher than a specified threshold (TH1) and/or lasts for a specified
   period of time (T1).

   When egress detects congestion level higher than TH1 and for a period
   of T1, it sends feedback signal to ingress periodically (T2) until

Wei                      Expires April 11, 2015                [Page 12]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   the congestion level is lower than TH1.

4.4 Congestion Control

   After ingress receives congestion information from egress, it will
   take actions to try to reduce the congestion. For example, ingress
   could choose to drop some packets or do certain traffic engineering

   Usually, network policy would have impact on what action is to be
   taken. For example, which packets to drop may be decided by the
   agreement between subscriber and network administrator. The specific
   choice of congestion alleviation measures taken by the ingress is out
   of scope of this document.

   The ingress will continue to implement control actions until there is
   no congestion feedback from the egress.

5. Congestion Feedback Protocol

   In different networks, there are always different tunnel protocols
   deployed. For instance, the congestion feedback can be done either by
   utilizing the existing tunnel protocol or using an alternative
   protocol. For example, in 3GPP network GTP (GPRS Tunnel
   Protocol)[TS29.060] is used as tunnel protocol to transmit traffic
   between network entities. And because GTP protocol is easy to be
   extended for additional information element, GTP itself would be a
   good choice for congestion feedback. In some other networks an
   independent protocol could be used for congestion feedback, for
   example the network using tunnel protocols such as IP-in-IP
   [RFC2003], GRE [RFC2784].

   Currently, this section mainly focuses on the discussion of
   independent protocols for congestion feedback. There are two choices
   for such an independent protocol, one is define as a new dedicated
   protocol from scratch, the other one is meant to evaluate and reuse
   the existing protocol(s).

5.1 Properties of Candidate Protocol

   To feedback congestion efficiently there are some properties that are
   desirable in the feedback protocol.

   1. Congestion friendliness. The feeding back traffics are coexistence
      with other traffics, so when congestion happens in the network,
      the feeding back traffic should be reduced, So that  feedback
      itself will not congest the network further when the network is

Wei                      Expires April 11, 2015                [Page 13]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

      already getting congested. In other words, feedback frequency
      should adjust to network's congestion level.

   2. Extensibility. The authors consider that using an existing
      protocol, or extensions to an existing protocol is preferable. The
      ability of a protocol to support modular extensions to report
      congestion level as feedback is a key attribute of the protocol
      under consideration.

   3. Compactness. In different situations, there may be different
      congestion information to be conveyed, and in order to reduce
      network load, the information to be conveyed should be selectable,
      i.e. only the required information should be possible to convey.

   4. In/Out of band signal. The feedback message could be along the
      same path with network data traffic, referred as in band signal;
      or go through a different path with network data traffic, referred
      as out of band signal.

5.2 IPFIX Extensions for Congestion Feedback

   This section outlines IPFIX extensions for feedback of congestion.
   The authors consider that IPFIX is a suitable protocol that is
   reasonably easy to extend to carry tunnel congestion reporting. The
   Feedback module acts as IPFIX exporter, and Collector module acts as
   IPFIX Collector.

   Since IPFIX is preferred to use SCTP as transport, it has the
   foundation for congestion-friendly behavior, and because SCTP allows
   partially reliable delivery [RFC3758] - IPFIX message channels can be
   tagged so that SCTP does not retransmit certain losses. This makes it
   safe during high levels of congestion in the reverse direction, to
   avoid a congestion collapse.. When congestion occurs in the network,
   the Exporter (Egress) can reduce the IPFIX traffic. Thus the feedback
   itself will not congest the network further when the network is
   already getting congested. When the Exporter detects network
   congestion, it can also reduce IPFIX traffic frequency to avoid more
   congestion in network while being able to sufficiently convey
   congestion status.

   Because the template mechanism in IPFIX is flexible, it allows the
   export of only the required information. Sending only the required
   information can also reduce network load.

Wei                      Expires April 11, 2015                [Page 14]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   The basic procedure for feedback using IPFIX is as follows:
   (1)The exporter inform the collector how to interpret the IEs in
      IPFIX message using template. Collector just accepts template
      passively; which IEs to send is configured by other means that not
      included in IPFIX specification.

   (2)The exporter meters the traffic and sends the congestion level to

   Congestion feedback using IPFIX is shown in the figures below. There
   are two variations to congestion feedback model using IPFIX. In the
   first one shown in Figure 6(a), congestion information is sent
   directly from egress to ingress and ingress makes decisions according
   this information. In the second case shown in Figure 6(b), congestion
   information is sent to a mediation controller instead of tunnel
   ingress; the controller is in charge of making decisions according to
   network congestion and control the behavior of ingress node, for
   example, reducing traffic or forbidding new traffic flows. In this
   model the congestion information from egress to controller is
   conveyed by IPFIX, but how controller controls the behavior of
   ingress is out of scope of this document.

          |                                         |
          |                                         |
          |                                         V
      +----------+         tunnel            +-----------+
      |Egress    |========================== |Inress     |
      |(Exporter)|                           |(Collector)|
      +----------+                           +-----------+

      (a) Direct Feedback.

Wei                      Expires April 11, 2015                [Page 15]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

          IPFIX    +-----------+
         --------->|Controller |#####################
         |         |(Collector)|                    #
         |         +-----------+                    #
         |                                          #
      +----------+          tunnel            +-----V-+
      |Egress    | ===========================|Ingress|
      |(Exporter)|                            +-------+

      (b) Mediated Feedback.

        Figure 6: IPFIX Congestion Feedback Models

   To support feeding back congestion information, some extensions to
   the IPFIX protocol are necessary.  According to the definition of
   congestion-related information defined in "Data Mode" section, new
   IEs conveying congestion level   is defined for IPFIX.

      Definition of new IE indicating congestion level.
         The congestion level calculated by exporter.
         Abstract Data Type: float32
         Data Type Semantics: quantity
         ElementId: TBD.
         Status: current

   The example below shows how IPFIX can be used for congestion

   (1) Sending Template Set The exporter use Template Set to inform the
   collector how to interpret the IEs in the following Data Set.

      |Set ID=2                |Length=n            |
      |Template ID=257         |Field Count=m       |
      |exporterIPv4Address=130 |Field Length=4      |
      |collectorIPv4Address=211|Field Length=4      |
      |CongestionLevel=TBD1    |Field Length=2      |
      |Enterprise Number=TBD2                       |

Wei                      Expires April 11, 2015                [Page 16]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   (2) Sending Data Set The exporter meters the traffic and sends the
   congestion information to collector by Data Set.

      |Set ID=257        |Length=n           |
      |                            |
      |                            |
      |0.1714                                |

       +--------+                            +---------+
       |Exporter|                            |Collector|
       +--------+                            +---------+
           |                                      |
           |                                      |
           |      (1)Sending Template Set         |
           |                                      |
      +--------+                                  |
      |metering|                                  |
      +--------+                                  |
           |     (2)Sending Data Set              |
           |               .                      |
           |               .                      |
           |               .                      |
           |                                      |
           |                                      |

        Figure 7: IPFIX Congestion Flow

   Before sending congestion information to collector, the exporter
   sends a Template set to Collector. The Template set specifies the
   structure and semantics of the subsequent Data Set containing
   congestion-related information. The Collector understands the Data
   Sets that follow according to Template Set that was sent previously.
   The exporting Process transmits the Template Set in advance of any
   Data Sets that use that Template ID, to help ensure that the
   Collector has the Template Record before receiving the first Data
   Record. Data Records that correspond to a Template Record may appear
   in the same and/or subsequent IPFIX Message(s).

Wei                      Expires April 11, 2015                [Page 17]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

   The Exporter meters the traffic passing through it and generates flow
   records. At this point, the Exporter may cache the records and then
   send congestion cumulative information to the collector. When
   Exporter detects that the network is heavily congested, it can change
   the feedback frequency to avoid adding more congestion to network.

   When receiving congestion related information, the Collector will
   make decisions to control the traffic entering the tunnel to reduce
   tunnel congestion.

5.3 Other Protocols

   A thorough evaluation of other protocols have not been performed at
   this time.

6. Benefits

   This section provides a short discussion about what benefits the
   tunnel congestion control would bring.

   Tunnel congestion control is a kind of local congestion control,
   where each tunnel is treated as an independent administrative domain
   in terms of congestion feedback and control, and it only responds to
   the congestion happened in the tunnel. The tunnel congestion control
   is complementary with e2e ECN control.

   The tunnel congestion feedback provides the network administrator
   with network congestion level information that can be used as an
   input for it local network management rather than relying solely on
   the e2e congestion control or blind traffic throttling. If the tunnel
   is congested it will be a waste of resource to allow new traffic to
   enter, because they may eventually get dropped in the tunnel. It's
   more efficient to have a control on new traffic at ingress.

7. Security Considerations

   This document describes the tunnel congestion calculation and
   feedback. For feeding back congestion, security mechanisms of IPFIX
   are expected to be sufficient. No additional security concerns are

8. IANA Considerations

   IANA assignment of parameters for IPFIX extension may need to be
   considered in this document.

Wei                      Expires April 11, 2015                [Page 18]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

9. References

9.1  Normative References

   [RFC2003]  Perkins, C., "IP Encapsulation within IP", RFC 2003,
              October 1996.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC2784]  Farinacci, D., Li, T., Hanks, S., Meyer, D., and P.
              Traina, "Generic Routing Encapsulation (GRE)", RFC 2784,
              March 2000.

   [RFC3168]  Ramakrishnan, K., Floyd, S., and D. Black, "The Addition
              of Explicit Congestion Notification (ECN) to IP",
              RFC 3168, September 2001.

   [RFC3758]  Stewart, R., Ramalho, M., Xie, Q., Tuexen, M., and P.
              Conrad, "Stream Control Transmission Protocol (SCTP)
              Partial Reliability Extension", RFC 3758, May 2004.

   [RFC4301]  Kent, S. and K. Seo, "Security Architecture for the
              Internet Protocol", RFC 4301, December 2005.

   [RFC5101]  Claise, B., Ed., "Specification of the IP Flow Information
              Export (IPFIX) Protocol for the Exchange of IP Traffic
              Flow Information", RFC 5101, January 2008.

   [RFC6040]  Briscoe, B., "Tunnelling of Explicit Congestion
              Notification", RFC 6040, November 2010.

   [I-D.boucadair-sfc-framework] Boucadair, M. etc, "Service Function
              Chaining: Framework & Architecture", draft-boucadair-sfc-
              framework-00(work in progress), October 2013.

   [I-D.zong-vnfpool-problem-statement] Zong, N. etc, "Virtualized
              Network Function (VNF) Pool Problem Statement", draft-
              zong-vnfpool-problem-statement-02(work in progress),
              January 2014.

9.2  Informative References

   [TS29.060]3GPP TS 29.060: "General Packet Radio Service (GPRS); GPRS
              Tunnelling Protocol (GTP) across the Gn and Gp interface".

Wei                      Expires April 11, 2015                [Page 19]

INTERNET DRAFT         Tunnel Congestion Feedback        October 8, 2014

Authors' Addresses

   Xinpeng Wei
   Beiqing Rd. Z-park No.156, Haidian District,
   Beijing,  100095, P. R. China

   Zhu Lei
   Beiqing Rd. Z-park No.156, Haidian District,
   Beijing,  100095, P. R. China

   Lingli Deng
   Beijing,  100095, P. R. China

Wei                      Expires April 11, 2015                [Page 20]