]> Google, Inc.

aldrin.ietf@gmail.com

Cisco Systems, Inc.

cpignata@cisco.com

Ericsson

gregory.mirsky@ericsson.com

Cisco Systems, Inc.

naikumar@cisco.com

BFD, seamless BFD, negotiation free, label verification, segment routing, IP This document describes various use cases for Seamless Bidirectional Forwarding Detection (S-BFD) and provides requirements such that protocol mechanisms allow for simplified detection of forwarding failures. These use cases support S-BFD, which is a simplified mechanism for using BFD with a large proportion of negotiation aspects eliminated, accelerating the establishment of a BFD session. The benefits of S-BFD include quick provisioning, as well as improved control and flexibility for network nodes initiating path monitoring.

Bidirectional Forwarding Detection (BFD), as defined in , is a lightweight protocol used to detect forwarding failures. Various protocols, applications, and clients rely on BFD for failure detection. Even though the protocol is lightweight and simple, there are certain use cases where faster setup of sessions and faster continuity checks of the data-forwarding paths are necessary. This document identifies these use cases and consequent requirements, such that enhancements and extensions result in a Seamless BFD (S-BFD) protocol. BFD is a simple and lightweight "Hello" protocol to detect data&nbhy;plane failures. With dynamic provisioning of forwarding paths on a large scale, establishing BFD sessions for each of those paths not only creates operational complexity but also causes undesirable delay in establishing or deleting sessions. The existing session establishment mechanism of the BFD protocol has to be enhanced in order to minimize the time for the session to come up to validate the forwarding path. This document specifically identifies various use cases and corresponding requirements in order to enhance BFD and other supporting protocols. Specifically, one key goal is removing the time delay (i.e., the "seam") between when a network node wants to perform a continuity test and when the node completes that continuity test. Consequently, "Seamless BFD" (S-BFD) has been chosen as the name for this mechanism. While the identified requirements could meet various use cases, it is outside the scope of this document to identify all of the possible and necessary requirements. Solutions related to the identified use cases and protocol&nbhy;specific enhancements or proposals are outside the scope of this document as well. Protocol definitions to support these use cases can be found in and .

The reader is expected to be familiar with the BFD , IP , MPLS , and Segment Routing terms and protocol constructs. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in .

BFD, as defined in , requires two network nodes to exchange locally allocated discriminators. These discriminators enable the identification of the sender and the receiver of BFD packets over the particular session. Subsequently, BFD performs proactive continuity monitoring of the forwarding path between the two. Several specifications describe BFD's multiple deployment uses: defines BFD over IPv4 and IPv6 for single IP hops. defines BFD over multi-hop paths. defines BFD for MPLS Label Switched Paths (LSPs). defines BFD for MPLS Pseudowires (PWs). Currently, BFD is best suited for verifying that two endpoints are mutually reachable or that an existing connection continues to be up and alive. In order for BFD to be able to initially verify that a connection is valid and that it connects the expected set of endpoints, it is necessary to provide each endpoint with the discriminators associated with the connection at each endpoint prior to initiating BFD sessions. The discriminators are used to verify that the connection is up and valid. Currently, the exchange of discriminators and the demultiplexing of the initial BFD packets are application dependent. If this information is already known to the endpoints of a potential BFD session, the initial handshake including an exchange of discriminators is unnecessary, and it is possible for the endpoints to begin BFD messaging seamlessly. A key objective of the S-BFD use cases described in this document is to avoid needing to exchange the initial packets before the BFD session can be established, with the goal of getting to the established state more quickly; in other words, the initial exchange of discriminator information is an unnecessary extra step that may be avoided for these cases. In a given scenario, an entity (such as an operator or a centralized controller) determines a set of network entities to which BFD sessions might need to be established. In traditional BFD, each of those network entities chooses a BFD Discriminator for each BFD session that the entity will participate in (see Section 6.3 of ). However, a key goal of S-BFD is to provide operational simplification. In this context, for S-BFD, each of those network entities is assigned one or more BFD Discriminators, and those network entities are allowed to use one Discriminator value for multiple sessions. Therefore, there may be only one or a few discriminators assigned to a node. These network entities will create an S-BFD listener session instance that listens for incoming BFD Control packets. When the mappings between specific network entities and their corresponding BFD Discriminators are known to other network nodes belonging to the same administrative domain, then, without having received any BFD packets from a particular target, a network entity in this network is able to send a BFD Control packet to the target's assigned discriminator in the Your Discriminator field. The target network node, upon reception of such a BFD Control packet, will transmit a response BFD Control packet back to the sender.

As per the BFD protocol , BFD sessions are established using a handshake mechanism prior to validating the forwarding path. This section outlines some use cases where the existing mechanism may not be able to satisfy the requirements identified. In addition, some of the use cases also stress the need for expedited BFD session establishment while preserving the benefits of forwarding failure detection using existing BFD mechanisms. Both of these high&nbhy;level goals result in the S-BFD use cases outlined in this document.

Even though bidirectional verification of forwarding paths is useful, there are scenarios where verification is only required in one direction between a pair of nodes. One such case is when a static route uses BFD to validate reachability to the next-hop IP router. In this case, the static route is established from one network entity to another. The requirement in this case is only to validate the forwarding path for that statically established unidirectional path. Validation of the forwarding path in the direction of the target entity to the originating entity is not required in this scenario. Many LSPs have the same unidirectional characteristics and unidirectional validation requirements. Such LSPs are common in Segment Routing and LDP&nbhy;based MPLS networks. A final example is when a unidirectional tunnel uses BFD to validate the reachability of an egress node. Additionally, validation of the unidirectional path has operational implications. If traditional BFD is to be used, the target network entity, as well as an initiator, has to be provisioned, even though reverse&nbhy;path validation with the BFD session is not required. However, in the case of unidirectional BFD, there is no need for provisioning on the target network entity -- only on the source entity. In this use case, a BFD session could be established in a single direction. When the target network entity receives the packet, it identifies the packet as BFD in an application-specific manner (for example, a destination UDP port number). Subsequently, the BFD module processes the packet, using the Your Discriminator value as context. Then, the network entity sends a response to the originator. This does not necessitate the requirement for establishment of a bidirectional session; hence, the two&nbhy;way handshake to exchange discriminators is not needed. The target node does not need to know the My Discriminator value of the source node. Thus, in this use case a requirement for BFD is to enable session establishment from the source network entity to the target network entity without the need to have a session (and state) for the reverse direction. Further, another requirement is that the BFD response from the target back to the sender can take any (in-band or out-of-band) path. The BFD module in the target network entity (for the BFD session), upon receipt of a BFD packet, starts processing the BFD packet based on the discriminator received. The source network entity can therefore establish a unidirectional BFD session without the bidirectional handshake and exchange of discriminators for session establishment.

In this use case, BFD is used to verify reachability before sending traffic via a path/LSP. This comes at a cost: traffic is prevented from using the path/LSP until BFD is able to validate reachability; this could take seconds due to BFD session bring&nbhy;up sequences , LSP Ping bootstrapping , etc. This use case would be better supported by eliminating the need for the initial BFD session negotiation. All it takes to be able to send BFD packets to a target and for the target to properly demultiplex these packets is for the source network entities to know what Discriminator values will be used for the session. This is also the case for S-BFD: the three-way handshake mechanism is eliminated during the bootstrapping of BFD sessions. However, this information is required at each entity to verify that BFD messages are being received from the expected endpoints; hence, the handshake mechanism serves no purpose. Elimination of the unnecessary handshake mechanism allows for faster reachability validation of BFD provisioned paths/LSPs. In addition, it is expected that some MPLS technologies will require traffic&nbhy;engineered LSPs to be created dynamically, perhaps driven by external applications, as, for example, in Software-Defined Networking (SDN). It will be desirable to perform BFD validation as soon as the LSPs are created, so as to use them. In order to support this use case, an S-BFD session is established without the need for session negotiation and exchange of discriminators.

Various technologies in the SDN domain that involve controller-based networks have evolved such that the intelligence, traditionally placed in a distributed and dynamic control plane, is separated from the networking entities themselves; instead, it resides in a (logically) centralized place. There are various controllers that perform the function of establishing forwarding paths for the data flow. Traffic engineering is one important function, where the path of the traffic flow is engineered, depending upon various attributes and constraints of the traffic paths as well as the network state. When the intelligence of the network resides in a centralized entity, the ability to manage and maintain the dynamic network, and its multiple data paths and node reachability, becomes a challenge. One way to ensure that the forwarding paths are valid and working is done by validation using BFD. When traffic&nbhy;engineered tunnels are created, it is operationally critical to ensure that the forwarding paths are working, prior to switching the traffic onto the engineered tunnels. In the absence of distributed control&nbhy;plane protocols, it may be desirable to verify any arbitrary forwarding path in the network. With tunnels being engineered by a centralized entity, when the network state changes, traffic has to be switched with minimum latency and without black-holing of the data. It is highly desirable in this centralized traffic-engineering use case that the traditional BFD session establishment and validation of the forwarding path do not become a bottleneck. If the controller or other centralized entity is able to very rapidly verify the forwarding path of a traffic&nbhy;engineered tunnel, it could steer the traffic onto the traffic&nbhy;engineered tunnel very quickly, thus minimizing adverse effects on a service. This is even more useful and necessary when the scale of the network and the number of traffic&nbhy;engineered tunnels grow. The cost associated with the time required for BFD session negotiation and establishment of BFD sessions to identify valid paths is very high when providing network redundancy is a critical issue.

A monitoring technique for a Segment Routing network based on a centralized controller is described in . Specific Operations, Administration, and Maintenance (OAM) requirements for Segment Routing are captured in . In validating this use case, one of the requirements is to ensure that the BFD packet's behavior is according to the monitoring specified for the segment and that the packet is U-turned at the expected node. This criterion ensures the continuity check to the adjacent Segment Identifier. To support this use case, the operational requirement is for BFD, initiated from a centralized controller, to perform liveness detection for any given segment in its domain.

When BFD sessions are being set up, torn down, or modified (i.e., when parameters such as intervals and multipliers are being modified), BFD requires additional packets, other than scheduled packet transmissions, to complete the negotiation procedures (i.e., Poll (P) bits and Final (F) bits; see Section 4.1 of ). There are scenarios where network resources are constrained: a node may require BFD to monitor a very large number of paths, or BFD may need to operate in low&nbhy;powered and traffic&nbhy;sensitive networks; these include microwave systems, low&nbhy;powered nanocells, and others. In these scenarios, it is desirable for BFD to slow down, speed up, stop, or resume at will and with a minimal number of additional BFD packets exchanged to modify the session or establish a new one. The established BFD session parameters, and such attributes as transmission interval and receiver interval, need to be modifiable without changing the state of the session.

The BFD protocol requires two endpoints to host BFD sessions, both sending packets to each other. This BFD model does not fit well with anycast address monitoring, as BFD packets transmitted from a network node to an anycast address will reach only one of potentially many network nodes hosting the anycast address. This use case verifies that a source node can send a packet to an anycast address and that the target node to which the packet is delivered can send a response packet to the source node. Traditional BFD cannot fulfill this requirement, since it does not provide for a set of BFD agents to collectively form one endpoint of a BFD session. The concept of a "target listener" in S-BFD fulfills this requirement. To support this use case, the BFD sender transmits BFD packets, which are received by any of the nodes hosting the anycast address to which the BFD packets are being sent. The anycast target that receives the BFD packet responds. This use case does not imply BFD session establishment with every node hosting the anycast address. Consequently, in this anycast use case, target nodes that do not happen to receive any of the BFD packets do not need to maintain any state, and the source node does not need to maintain separate state for each target node.

BFD for multi-hop paths and BFD for MPLS LSPs perform end-to-end validation, traversing multiple network nodes. BFD has been designed to declare a failure to receive some number of consecutive packets. This failure can be caused by a fault anywhere along these paths. Fast failure detection allows for rapid fault detection and consequent rapid path recovery procedures. However, operators often have to follow up, manually or automatically, to attempt to identify and localize the fault that caused BFD sessions to fail (i.e., fault isolation). If Equal-Cost Multipath (ECMP) is used, the usage of other tools to isolate the fault (e.g., traceroute) may cause the packets to traverse a different path through the network. In addition, the longer it takes from the time of BFD session failure to the time that fault isolation begins, the more likely the fault will not be isolated (e.g., a fault may be corrected via rerouting or some other means during that time). If BFD had built&nbhy;in fault-isolation capability, fault isolation would be triggered when the fault was first detected. This embedded fault isolation would be more effective (i.e., faults would be detected sooner) if those BFD fault-isolation packets were load-balanced in the same way as the BFD packets that were dropped. This use case describes S-BFD fault-isolation capabilities, utilizing a TTL field (e.g., as described in Section 5.1.1 of ) or using fields that indicate status.

BFD is capable of providing very fast failure detection, as relevant network nodes continuously transmit BFD packets at the negotiated rate. If BFD packet transmission is interrupted, even for a very short period of time, BFD can declare a failure irrespective of path liveness. On a system where BFD is running, it is possible for certain events to (intentionally or unintentionally) cause a brief interruption of BFD packet transmissions. With distributed architectures of BFD implementations, this case can be prevented. This use case is for an S-BFD node running multiple BFD sessions to the same target node, with those sessions hosted on different system modules (e.g., in different CPU instances). This can reduce false failures, resulting in a more stable network. To support this use case, a mapping between the multiple discriminators on a single system and the specific entity within that system is required.

BFD for MPLS LSPs, defined in , describes procedures for running BFD as an LSP in-band continuity check mechanism by using MPLS Echo Request messages to bootstrap the BFD session on the target (i.e., egress) node. Section 4 of also describes the possibility of running multiple BFD sessions per alternative of LSPs. further clarifies the procedures, for both ingress and egress nodes, regarding how to bootstrap, maintain, and remove multiple BFD sessions for the same <MPLS LSP, FEC> tuple ("FEC" means Forwarding Equivalence Class). However, this mechanism still requires the use of MPLS LSP Ping for bootstrapping, round trips for initialization, and keeping state at the receiver. In the presence of ECMP within an MPLS LSP, it may be desirable to run in-band monitoring that exercises every path of this ECMP. Otherwise, there will be scenarios where an in-band BFD session remains up through one path but traffic is black-holing over another path. A BFD session per ECMP path of an LSP requires the definition of procedures that update in terms of how to bootstrap and maintain the correct set of BFD sessions on the egress node. However, for traditional BFD, that requires the constant use of MPLS Echo Request messages to create and delete BFD sessions on the egress node when ECMP paths and/or corresponding load-balance hash keys change. If a BFD session over any paths of the LSP can be instantiated, stopped, and resumed without requiring additional procedures for bootstrapping via an MPLS Echo Request message, it would greatly simplify both implementations and operations and would benefit network devices, as less processing would be required by them. To support this requirement, multiple S-BFD sessions need to be established over different ECMP paths between the same pair of source and target nodes.

Upon receipt of an S-BFD packet, a target network entity (for the S-BFD session) MUST process the packet based on the discriminator received in the BFD packet. If the S-BFD context is found, the target network entity MUST be able to send a response. The source network entity MUST be able to establish a unidirectional S-BFD session without the bidirectional handshake of discriminators for session establishment. The S-BFD session MUST be able to be established without the need for the exchange of discriminators during session negotiation. In a Segment Routed network, S-BFD MUST be able to perform liveness detection initiated from a centralized controller for any given segment in its domain. The established S-BFD session parameters and attributes, such as transmission interval and reception interval, MUST be modifiable without changing the state of the session. An S-BFD source network entity MUST be able to send Control packets to an anycast address. These packets are received and processed by any node hosting the anycast address. The S-BFD entity MUST be able to receive responses to S-BFD Control packets from any of these anycast nodes, without establishing a separate S-BFD session with every node hosting the anycast address. S-BFD SHOULD support fault-isolation capability, which MAY be triggered when a fault is encountered. S-BFD SHOULD be able to establish multiple sessions between the same pair of source and target nodes. This requirement enables but does not guarantee the ability to monitor divergent paths in ECMP environments. It also provides resiliency in distributed router architectures. The mapping between BFD Discriminators and particular entities (e.g., ECMP paths, line cards) is out of scope for the S-BFD protocol. The S-BFD protocol MUST provide mechanisms for loop detection and prevention, protecting against malicious attacks attempting to create packet loops. S-BFD MUST incorporate robust security protections against impersonators, malicious actors, and various active and passive attacks. The simple and accelerated establishment of an S-BFD session should not negatively affect security.

This document details use cases for S-BFD and identifies various associated requirements. Some of these requirements are security related. The use cases described herein do not expose a system to abuse or additional security risks. Since some negotiation aspects are eliminated, a misconfiguration can result in S-BFD packets being sent to an incorrect node. If this receiving node runs S-BFD, the packet will be discarded due to discriminator mismatch. If the node does not run S-BFD, the packet will be naturally discarded. The proposed new protocols, extensions, and enhancements for S-BFD supporting these use cases and realizing these requirements will address associated security considerations. S-BFD should not have reduced security capabilities as compared to traditional BFD.

&RFC2119; &RFC5881; &RFC5883; &RFC5880; &RFC5884; &RFC5885; &RFC2460; &RFC3031; &RFC4379; &RFC7726;

The authors would like to thank Tobias Gondrom and Eric Gray for their insightful and useful comments. The authors appreciate the thorough review and comments provided by Dale R. Worley.

The following are key contributors to this document: Manav Bhatia, Ionos Networks Satoru Matsushima, Softbank Glenn Hayden, ATT Santosh P K Mach Chen, Huawei Nobo Akiya, Big Switch Networks