]> Aalyria Technologies, Inc.

ek.ietf@gmail.com

Internet Delay/Disruption Tolerant Networking Delay and Disruption Tolerant Networking DTN Bundle Protocol BP BTPU Ethernet This document specifies the use of the Bundle Transfer Protocol - Unidirectional (BTPU) as a Convergence Layer directly over Ethernet, and requests allocation of an EtherType and a multicast MAC address for that purpose. This provides an alternative to IP-based convergence layers for environments where Ethernet forwarding is operationally feasible but IP routing is unavailable or operationally undesirable. About This Document The latest revision of this draft can be found at . Status information for this document may be found at . Discussion of this document takes place on the Delay/Disruption Tolerant Networking Working Group mailing list (), which is archived at . Subscribe at . Source for this draft and an issue tracker can be found at .

Introduction This document specifies how BTPU is carried directly in Ethernet frames, enabling its use as a Convergence Layer for environments where Bundle Protocol nodes are connected by Ethernet or Ethernet-like technologies. It defines the encapsulation (), the mapping of BTPU's logical channel onto Ethernet addressing, and Ethernet-specific operational and security considerations. To support this, the following Ethernet parameters are requested:

• an EtherType to identify frames carrying BTPU payloads ()
• a multicast MAC address for transmission to receivers whose unicast MAC addresses are not yet known ()

This convergence layer is applicable to:

• physical Ethernet LANs;
• Layer 2 services that present an Ethernet segment to Bundle Protocol Agents, such as overlay networks, cloud-hosted virtual networks, or Ground-Station-as-a-Service (GSaaS) infrastructure, provided the service carries arbitrary EtherTypes (and, where the group MAC address is to be used, non-IP multicast frames); and
• technologies supporting Ethernet framing, e.g., DVB-GSE (), the 3GPP 5G Ethernet PDU Session type (Section 5.6.10.2 of ), and the US Space Development Agency's Optical Communications Terminal standard (Section 3.4.8 of ).

Primary use cases include mission modeling, testbed environments, and deployments where IP routing is unavailable or adds unnecessary complexity.

Conventions and Definitions 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 BCP 14 when, and only when, they appear in all capitals, as shown here.

Encapsulation A BTPU Link-layer PDU (Section 2.1 of ) is carried as the payload of an Ethernet frame whose EtherType field is set to the value assigned in . The Link-layer PDU consists of the octets following the EtherType field (or following the last tag, if one or more 802.1Q tags are present) up to, but not including, the Frame Check Sequence (FCS). Its contents are a sequence of BTPU Messages as defined in .

BTPU over Ethernet frame format (shown with one 802.1Q tag)

Minimum Frame Size IEEE 802.3 requires a minimum frame size of 64 octets. A frame whose payload is shorter than 46 octets (42 octets when an 802.1Q tag is present) is extended by the transmitting MAC with a PAD field whose contents are unspecified by , although in practice they are conventionally zero-filled. Because EtherType-framed Ethernet carries no payload length indication, a receiver cannot distinguish such MAC padding from BTPU payload. To avoid relying on MAC padding behavior, senders SHOULD ensure that every Link-layer PDU is at least 46 octets long by appending a Definite Padding Message (Section 8.5 of ) as needed. Receivers MUST tolerate trailing octets that do not parse as a BTPU Message in frames of minimum size, and SHOULD treat trailing zero octets as an Indefinite Padding Message (Section 8.6 of ). Beyond the minimum frame size, Ethernet frames are variable length and no further padding is required.

Applicability and Limitations

BTPU Protocol Compliance This document specifies Ethernet encapsulation for . All protocol requirements, features, and recommendations defined in apply to this Ethernet profile. Because Bundles commonly exceed the Ethernet MTU (), implementations MUST support BTPU segmentation (Section 4 of ) for both transmission and reception. BTPU's 20-bit Message Length field does not constrain Ethernet use, as it comfortably exceeds any Ethernet frame size.

BTPU Logical Channels operates over a logical channel between a sender and one or more receivers; each logical channel is an independent instance of the protocol with its own Transfer Number sequence and Transfer Window (Sections 4 and 5 of ). For Ethernet, the logical channel is identified by the tuple of:

• the interface on which the frame is transmitted or received, where each VLAN, as identified by the VLAN Identifier(s) of any 802.1Q tag(s) present, constitutes a distinct interface;
• the source MAC address; and
• the destination MAC address.

Each unique combination defines a separate logical channel. In particular, frames from the same sender addressed to a peer's unicast MAC address and frames addressed to the group MAC address () belong to different logical channels, and a receiver maintains independent BTPU state for each. The Priority Code Point (PCP) and Drop Eligible Indicator (DEI) fields of an 802.1Q tag do not identify the logical channel. Because a single Link-layer PDU can carry Messages belonging to several Transfers of differing priority, this document makes no recommendation about how those fields are set. The logical channel governs the sequencing and windowing of segmented Transfers. Bundle Messages (Section 8.1 of ) carry no Transfer Number and require no per-channel state at the receiver beyond that needed to identify the transmitting node. The Transfer Window size is configured out of band (Section 5 of ). Absent such configuration, implementations SHOULD use the default recommended by . Note that a receiver whose Transfer Window is smaller than the sender's will prematurely discard in-progress Transfers; this is of particular concern for channels using the group MAC address, where a single sender's frames are processed by many independently configured receivers. Technologies that carry Ethernet-framed payloads and use this EtherType () but lack the above channel identifiers need to define an equivalent logical channel identifier, e.g. from technology-specific source and destination identifiers and any protocol-specific channel discriminators. Such definitions are outside the scope of this document.

Use of the Group MAC Address A sender that does not know the unicast MAC address of the intended next-hop node MAY transmit BTPU frames to the group MAC address assigned in . All BTPU receivers in the broadcast domain will receive such frames; from the perspective of each receiver they share a single logical channel identified by the sender's source MAC address and the group destination address (). A node that receives a bundle in a frame addressed to the group MAC address processes it as follows:

• If the bundle's destination EID identifies an endpoint of which the receiving node is a member, the bundle is delivered as specified in .
• Otherwise, if the receiving node has been explicitly configured to act as a next hop for the transmitting node (as identified by the frame's source MAC address and logical channel), the bundle is forwarded as specified in . Implementations MUST NOT enable such forwarding by default.
• Otherwise, the receiving node MUST discard the bundle without further processing. In particular, it MUST NOT forward the bundle and MUST NOT generate any status report concerning it; the bundle is treated as though it had never been received.

Without this rule, a single frame carrying a bundle addressed to a singleton endpoint would be forwarded by every node in the broadcast domain that is not the destination, producing as many copies as there are receivers (see ). This mirrors the prohibition on forwarding link-layer broadcasts that carry unicast destinations in Section 5.3.4 of . A sender learns the unicast MAC address of a peer either through configuration or by observing the source MAC address of BTPU frames received from that peer; the latter requires the peer to transmit, which BTPU does not itself guarantee. Once a peer's unicast MAC address is known, senders SHOULD transmit to it by unicast so as to avoid unnecessary processing by other nodes and unnecessary flooding by switches, which cannot prune non-IP multicast traffic.

Congestion Control BTPU provides no congestion control and assumes that the sending rate is managed by a mechanism outside the protocol (Section 11 of ). Ethernet offers only hop-by-hop flow control, namely PAUSE frames (Clause 31 of ) and Priority-based Flow Control (). These mechanisms act on a single link rather than end to end across a bridged network, may be disabled by operators to avoid head-of-line blocking of unrelated traffic, and are ineffective over links whose delay is large relative to available buffering. They are therefore not a substitute for congestion control. Consistent with Section 11 of , BTPU over Ethernet MUST NOT be deployed on a segment where congestion can occur unless the sending rate is bounded by an external mechanism, such as static rate limiting or a schedule agreed among the nodes sharing the segment. Where no such mechanism is available, a convergence layer providing congestion control (e.g., ) is recommended instead.

Relationship to IP-based Convergence Layers IP-based convergence layers (TCPCL , UDPCLv2 ) remain recommended where IP infrastructure exists. This Ethernet convergence layer addresses scenarios where:

• no operational IP addressing or routing is available
• only link-local IP addresses are available and no peer discovery mechanism is deployed
• direct Ethernet operation simplifies deployment and management

Direct Ethernet encapsulation also avoids the 28 octets (IPv4/UDP) or 48 octets (IPv6/UDP) of network- and transport-layer headers per frame, plus any convergence-layer framing. In many deployments this saving is minor, but it can be significant where the usable payload per frame is small, where link capacity is scarce, or where eliminating IP and UDP header processing simplifies line-rate hardware implementation. The primary motivation for this specification, however, remains operation in environments where IP is unavailable or undesirable.

Assignment Considerations This document requests one EtherType from the IEEE Registration Authority and one multicast MAC address from IANA, as described below.

IEEE Assignment Considerations

EtherType Following the procedure in Section 5.5 of : the IESG is requested to approve applying to the IEEE Registration Authority for an EtherType for BTPU. (The IESG should communicate its approval to IANA and to those concerned with this document. IANA will forward the IESG Approval to the registry expert of the "EtherType" registry from the "IEEE 802 Numbers" registry group who will make the application to the IEEE Registration Authority, keeping IANA informed.) Upon assignment, IANA is requested to record the following entry in the "EtherType" registry of the "IEEE 802 Numbers" registry group : EtherType registry entry

Ethertype (decimal)	Ethertype (hex)	Description	Reference
TBD-ET-DEC	TBD-ET	Bundle Transfer Protocol - Unidirectional (BTPU)	RFC XXXX

[RFC Editor: please replace TBD-ET-DEC and TBD-ET throughout this document with the assigned EtherType, and RFC XXXX with this document's RFC number; then remove this note.]

IANA Considerations

Multicast MAC Address IANA is requested to assign one multicast EUI-48 identifier under the IANA OUI, from the block used for very small assignments (01-00-5E-90-00-00 through 01-00-5E-90-00-FF; Section 2.1.3 of ), and to record it in the "IANA Multicast 48-bit MAC Addresses" registry with this document as the reference. Per Section 2.1.5 of , this assignment is subject to Expert Review. The address allows BTPU senders to reach all BTPU receivers within a broadcast domain without prior knowledge of their individual unicast MAC addresses, as described in . A dedicated group address is requested, rather than use of the broadcast address, so that stations not participating in BTPU can filter these frames in hardware. [RFC Editor: please replace TBD-MAC below with the assigned address and remove this note.] Assigned address: TBD-MAC The completed template of Appendix A.1 of follows:

Applicant Name:

IETF DTN Working Group

Applicant Email:

dtn@ietf.org

Applicant Telephone:

(to be supplied at submission)

Use Name:

Bundle Transfer Protocol - Unidirectional (BTPU) over Ethernet

Document:

RFC XXXX (this document)

EUI-48 or EUI-64:

EUI-48

Size of Block requested:

1 (2**0)

Multicast, unicast, or both:

Multicast

Operational Considerations

Checksums As noted in Section 3.5 of , the Bundle Protocol assumes that Bundles are transmitted over an erasure channel, i.e., one that either delivers a PDU correctly or not at all. The Ethernet Frame Check Sequence (FCS) provides this property for a single link. However, the FCS is verified and regenerated at each bridge, so it does not detect corruption that occurs within a bridge, and the error-detection strength of its 32-bit CRC diminishes as frame length increases. Deployments requiring end-to-end integrity assurance SHOULD rely on Bundle-layer mechanisms such as block CRCs () or BPSec Block Integrity Blocks () rather than on the FCS alone.

MTU and Jumbo Frames Implementations MUST support transmission and reception of Link-layer PDUs of up to 1500 octets, the maximum payload size of a basic Ethernet frame . This limit applies to the BTPU Link-layer PDU itself; 802.1Q tags, if present, do not reduce it. Implementations MAY support non-standard "jumbo" frames (commonly with payloads of up to 9000 octets), but SHOULD do so only when explicitly configured by an operator who has verified that every link and bridge on the path supports the larger size. BTPU has no path MTU discovery mechanism, and Ethernet bridges silently discard frames that exceed their supported size. Implementations therefore SHOULD default to a 1500-octet Link-layer PDU.

Security Considerations The security considerations of and apply. BTPU itself provides no security mechanisms and relies on lower and/or upper layers (Section 10 of ). The following considerations are specific to carrying BTPU over Ethernet.

Denial of Service BTPU assumes the sending rate is controlled by a mechanism out of scope for the protocol and has no built-in mechanism for identifying or mitigating any congestion a sender might cause (). Use of this protocol on some networks, a shared LAN segment for example, may cause a Denial-of-Service by flooding Ethernet switches and stations.

Multicast Amplification A single frame sent to the group MAC address () is delivered to every BTPU receiver in the broadcast domain. If receivers were to process such bundles exactly as they process unicast-received bundles, an attacker (or a misconfigured node) could obtain significant amplification from a single frame:

• each receiver that is not the bundle's destination would attempt to forward the bundle toward that destination, producing as many copies as there are receivers; and
• each receiver that subsequently deletes the bundle could emit a status report to the bundle's report-to endpoint, if one was requested, producing as many status reports as there are receivers.

The processing rules in are intended to prevent both forms of amplification: bundles received via the group MAC address are forwarded only by nodes explicitly configured as a next hop for the transmitting node, and are otherwise discarded without generating any status report. Implementations MUST NOT enable such forwarding by default. Operators SHOULD limit the number of nodes configured to forward on behalf of any given transmitting node.

Frame Injection and Spoofing Ethernet provides no authentication of the source MAC address. An attacker with access to the link can observe the source and destination MAC addresses and Transfer Numbers in use on a logical channel and inject frames with a forged source address. Such an attacker can corrupt an in-progress Transfer by injecting a Transfer Segment Message, abort it by injecting a Transfer Cancel Message, or cause the receiver to discard all in-progress Transfers from that sender by injecting a Message with a Transfer Number far ahead of the current window (Section 5 of ). Corrupted bundles can be detected by Bundle-layer integrity mechanisms (), but the transfers themselves are lost. Any attacker with access to the link, or with sufficient knowledge of local Bundle forwarding configuration so as to inject BTPU frames and cause them to be sent to an Ethernet peer, may also overwhelm the receiver to the point of Denial of Service to other on-link senders. These attacks are mitigated by restricting link access and authenticating frame origin using the mechanisms described in , or by architectural properties of the link that exclude untrusted stations.

Link-layer Security IEEE standards include several security mechanisms that may be used in Ethernet networks. Examples of Ethernet-level security mechanisms a network might deploy include IEEE 802.1X (), which may be used to restrict access to the link to authorized participants, and IEEE 802.1AE (), which provides origin authentication and integrity for each frame and, optionally, confidentiality of the entire BTPU payload. IEEE 802.1AE can also provide replay protection, but only when configured to do so and only for frames outside the configured replay window; deployments relying on it for this purpose need to enable replay protection with a window sized appropriately for the link. In some deployments, a link may be considered secure against on-link attackers by virtue of its architecture, e.g. in cloud networking configurations where access to the virtual link is the responsibility of cloud security functions.

Packet Reordering, Duplication, and Replay Packet reordering and duplication are handled by the BTPU protocol. However, an on-link attacker may replay traffic to effectively repeat a Bundle transfer. Even if a link can be made secure (), repeat delivery of specific Bundles may happen for other reasons. Duplicate bundles can be detected by the Bundle Protocol Agent using the bundle identifier (Section 4.2.4 of ); whether and for how long to retain such state is a matter of BPA policy and is independent of the convergence layer.

Filtering A common security paradigm is to "default deny" all traffic patterns that, broadly, do not conform to operator expectations. In such environments the BTPU EtherType and, where used, the group MAC address () need to be explicitly permitted on a given Ethernet segment before BTPU Messages can be successfully transmitted.

References Normative References This document presents a specification for the Bundle Protocol, adapted from the experimental Bundle Protocol specification developed by the Delay-Tolerant Networking Research Group of the Internet Research Task Force and documented in RFC 5050. Aalyria Technologies This document defines a protocol for the unidirectional transfer of large binary objects, typically Bundle Protocol version 7 bundles, between two nodes connected by a unidirectional, unreliable, frame- based link-layer protocol, without requiring IP services. The protocol does not require a return path for acknowledgements, but instead supports data repetition as a mechanism to protect against data loss. It fully supports the disaggregation of flows of binary objects of different priority, preventing head-of-line blocking impacting performance. The wire format of the protocol is designed to enable performant implementation in hardware or software, with the aim of enabling protocol implementations to run at the line-rate of the underlying link-layer protocol. Some IETF protocols make use of Ethernet frame formats and IEEE 802 parameters. This document discusses several aspects of such parameters and their use in IETF protocols, specifies IANA considerations for assignment of points under the IANA Organizationally Unique Identifier (OUI), and provides some values for use in documentation. This document obsoletes RFC 7042. IEEE IEEE In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. Informative References This document specifies the preferred method for transporting Delay- and Disruption-Tolerant Networking (DTN) protocol data over the Internet using datagrams. It covers convergence layers for the Bundle Protocol (RFC 5050), as well as the transportation of segments using the Licklider Transmission Protocol (LTP) (RFC 5326). UDP and the Datagram Congestion Control Protocol (DCCP) are the candidate datagram protocols discussed. UDP can only be used on a local network or in cases where the DTN node implements explicit congestion control. DCCP addresses the congestion control problem, and its use is recommended whenever possible. This document is a product of the Delay-Tolerant Networking Research Group (DTNRG) and represents the consensus of the DTNRG. The Johns Hopkins University Applied Physics Laboratory Science Applications International Corporation This document describes a UDP convergence layer (UDPCL) for Delay- Tolerant Networking (DTN). This version of the UDPCL protocol clarifies requirements of the earlier experimental RFC 7122, adds discussion of multicast addressing, congestion signaling, and updates to the Bundle Protocol (BP) contents, encodings, and convergence layer requirements in BP version 7. Specifically, the UDPCL uses CBOR-encoded BPv7 bundles as its service data unit being transported and provides an unacknowledged transport of such bundles. This version of UDPCL also includes security and extensibility mechanisms. This document describes a TCP convergence layer (TCPCL) for Delay-Tolerant Networking (DTN). This version of the TCPCL protocol resolves implementation issues in the earlier TCPCL version 3 as defined in RFC 7242 and provides updates to the Bundle Protocol (BP) contents, encodings, and convergence-layer requirements in BP version 7 (BPv7). Specifically, TCPCLv4 uses BPv7 bundles encoded by the Concise Binary Object Representation (CBOR) as its service data unit being transported and provides a reliable transport of such bundles. This TCPCL version also includes security and extensibility mechanisms. This memo defines and discusses requirements for devices that perform the network layer forwarding function of the Internet protocol suite. [STANDARDS-TRACK] This document defines a security protocol providing data integrity and confidentiality services for the Bundle Protocol (BP). IANA n.d. IEEE IEEE ETSI 3GPP n.d. Space Development Agency

Acknowledgments Thanks to Wes Eddy, Jörg Ott, Brian Sipos, and Rick Taylor for numerous discussions and contributions.