Internet-Draft | DTN TCPCLv4 | November 2020 |
Sipos, et al. | Expires 18 May 2021 | [Page] |
This document describes a TCP-based convergence layer (TCPCL) for Delay-Tolerant Networking (DTN). This version of the TCPCL protocol resolves implementation issues in the earlier TCPCL Version 3 of RFC7242 and updates to the Bundle Protocol (BP) contents, encodings, and convergence layer requirements in BP Version 7. Specifically, the TCPCLv4 uses CBOR-encoded BPv7 bundles as its service data unit being transported and provides a reliable transport of such bundles. This version of TCPCL also includes security and extensibility mechanisms.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
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."¶
This Internet-Draft will expire on 18 May 2021.¶
Copyright (c) 2020 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 (https://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.¶
This document describes the TCP-based convergence-layer protocol for Delay-Tolerant Networking. Delay-Tolerant Networking is an end-to-end architecture providing communications in and/or through highly stressed environments, including those with intermittent connectivity, long and/or variable delays, and high bit error rates. More detailed descriptions of the rationale and capabilities of these networks can be found in "Delay-Tolerant Network Architecture" [RFC4838].¶
An important goal of the DTN architecture is to accommodate a wide range of networking technologies and environments. The protocol used for DTN communications is the Bundle Protocol Version 7 (BPv7) [I-D.ietf-dtn-bpbis], an application-layer protocol that is used to construct a store-and-forward overlay network. BPv7 requires the services of a "convergence-layer adapter" (CLA) to send and receive bundles using the service of some "native" link, network, or Internet protocol. This document describes one such convergence-layer adapter that uses the well-known Transmission Control Protocol (TCP). This convergence layer is referred to as TCP Convergence Layer Version 4 (TCPCLv4). For the remainder of this document, the abbreviation "BP" without the version suffix refers to BPv7. For the remainder of this document, the abbreviation "TCPCL" without the version suffix refers to TCPCLv4.¶
The locations of the TCPCL and the BP in the Internet model protocol stack (described in [RFC1122]) are shown in Figure 1. In particular, when BP is using TCP as its bearer with TCPCL as its convergence layer, both BP and TCPCL reside at the application layer of the Internet model.¶
+-------------------------+ | DTN Application | -\ +-------------------------| | | Bundle Protocol (BP) | -> Application Layer +-------------------------+ | | TCP Conv. Layer (TCPCL) | | +-------------------------+ | | TLS (optional) | -/ +-------------------------+ | TCP | ---> Transport Layer +-------------------------+ | IPv4/IPv6 | ---> Network Layer +-------------------------+ | Link-Layer Protocol | ---> Link Layer +-------------------------+
This document describes the format of the protocol data units passed between entities participating in TCPCL communications. This document does not address:¶
Any TCPCL implementation requires a BP agent to perform those above listed functions in order to perform end-to-end bundle delivery.¶
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 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This section contains definitions specific to the TCPCL protocol.¶
This is the notional TCPCL application that initiates TCPCL sessions. This design, implementation, configuration, and specific behavior of such an entity is outside of the scope of this document. However, the concept of an entity has utility within the scope of this document as the container and initiator of TCPCL sessions. The relationship between a TCPCL entity and TCPCL sessions is defined as follows:¶
These relationships are illustrated in Figure 2. For most TCPCL behavior within a session, the two entities are symmetric and there is no protocol distinction between them. Some specific behavior, particularly during session establishment, distinguishes between the active entity and the passive entity. For the remainder of this document, the term "entity" without the prefix "TCPCL" refers to a TCPCL entity.¶
The relationship between connections, sessions, and streams is shown in Figure 3.¶
+--------------------------------------------+ | TCPCL Entity | | | +----------------+ | +--------------------------------+ | | |-+ | | Actively Initiated Session #1 +------------->| Other | | | +--------------------------------+ | | TCPCL Entity's | | | ... | | Passive | | | +--------------------------------+ | | Listener | | | | Actively Initiated Session #n +------------->| | | | +--------------------------------+ | +----------------+ | | | +-----------------+ | +---------------------------+ | | +---| +---------------------------+ | +----------------+ | | | | Optional Passive | | | |-+ | | +-| Listener(s) +<-------------+ | | | | +---------------------------+ | | | | | | | | Other | | | | +---------------------------------+ | | TCPCL Entity's | | | +--->| Passively Initiated Session #1 +-------->| Active | | | | +---------------------------------+ | | Initiator(s) | | | | | | | | | | +---------------------------------+ | | | | | +--->| Passively Initiated Session #n +-------->| | | | +---------------------------------+ | +----------------+ | | | +-----------------+ +--------------------------------------------+
+---------------------------+ +---------------------------+ | "Own" TCPCL Session | | "Other" TCPCL Session | | | | | | +----------------------+ | | +----------------------+ | | | TCP Connection | | | | TCP Connection | | | | | | | | | | | | +-----------------+ | | Messages | | +-----------------+ | | | | | Own Inbound | +--------------------+ | Peer Outbound | | | | | | Transfer Stream | | Transfer Stream | | | | | | ----- |<---[Seg]--[Seg]--[Seg]---| ----- | | | | | | RECEIVER |---[Ack]----[Ack]-------->| SENDER | | | | | +-----------------+ +-----------------+ | | | | | | | | +-----------------+ +-----------------+ | | | | | Own Outbound |-------[Seg]---[Seg]----->| Peer Inbound | | | | | | Transfer Stream |<---[Ack]----[Ack]-[Ack]--| Transfer Stream | | | | | | ----- | | ----- | | | | | | SENDER | +--------------------+ | RECEIVER | | | | | +-----------------+ | | | | +-----------------+ | | | +-----------------------+ | | +---------------------+ | +----------------------------+ +--------------------------+
The service of this protocol is the transmission of DTN bundles via the Transmission Control Protocol (TCP). This document specifies the encapsulation of bundles, procedures for TCP setup and teardown, and a set of messages and entity requirements. The general operation of the protocol is as follows.¶
This version of the TCPCL provides the following services to support the overlaying Bundle Protocol agent. In all cases, this is not an API definition but a logical description of how the CL can interact with the BP agent. Each of these interactions can be associated with any number of additional metadata items as necessary to support the operation of the CL or BP agent.¶
The TCPCL entity indicates to the BP agent when the session state changes. The top-level session states indicated are:¶
First, one entity establishes a TCPCL session to the other by initiating a TCP connection in accordance with [RFC0793]. After setup of the TCP connection is complete, an initial Contact Header is exchanged in both directions to establish a shared TCPCL version and negotiate the use of TLS security (as described in Section 4). Once contact negotiation is complete, TCPCL messaging is available and the session negotiation is used to set parameters of the TCPCL session. One of these parameters is a Node ID that each TCPCL Entity is acting as. This is used to assist in routing and forwarding messages by the BP Agent and is part of the authentication capability provided by TLS.¶
Once negotiated, the parameters of a TCPCL session cannot change and if there is a desire by either peer to transfer data under different parameters then a new session must be established. This makes CL logic simpler but relies on the assumption that establishing a TCP connection is lightweight enough that TCP connection overhead is negligible compared to TCPCL data sizes.¶
Once the TCPCL session is established and configured in this way, bundles can be transferred in either direction. Each transfer is performed by segmenting the transfer data into one or more XFER_SEGMENT messages. Multiple bundles can be transmitted consecutively in a single direction on a single TCPCL connection. Segments from different bundles are never interleaved. Bundle interleaving can be accomplished by fragmentation at the BP layer or by establishing multiple TCPCL sessions between the same peers. There is no fundamental limit on the number of TCPCL sessions which a single entity can establish beyond the limit imposed by the number of available (ephemeral) TCP ports of the active entity.¶
A feature of this protocol is for the receiving entity to send acknowledgment (XFER_ACK) messages as bundle data segments arrive. The rationale behind these acknowledgments is to enable the transmitting entity to determine how much of the bundle has been received, so that in case the session is interrupted, it can perform reactive fragmentation to avoid re-sending the already transmitted part of the bundle. In addition, there is no explicit flow control on the TCPCL layer.¶
A TCPCL receiver can interrupt the transmission of a bundle at any point in time by replying with a XFER_REFUSE message, which causes the sender to stop transmission of the associated bundle (if it hasn't already finished transmission). Note: This enables a cross-layer optimization in that it allows a receiver that detects that it already has received a certain bundle to interrupt transmission as early as possible and thus save transmission capacity for other bundles.¶
For sessions that are idle, a KEEPALIVE message is sent at a negotiated interval. This is used to convey entity live-ness information during otherwise message-less time intervals.¶
A SESS_TERM message is used to initiate the ending of a TCPCL session (see Section 6.1). During termination sequencing, in-progress transfers can be completed but no new transfers can be initiated. A SESS_TERM message can also be used to refuse a session setup by a peer (see Section 4.3). Regardless of the reason, session termination is initiated by one of the entities and responded-to by the other as illustrated by Figure 13 and Figure 14. Even when there are no transfers queued or in-progress, the session termination procedure allows each entity to distinguish between a clean end to a session and the TCP connection being closed because of some underlying network issue.¶
Once a session is established, TCPCL is a symmetric protocol between the peers. Both sides can start sending data segments in a session, and one side's bundle transfer does not have to complete before the other side can start sending data segments on its own. Hence, the protocol allows for a bi-directional mode of communication. Note that in the case of concurrent bidirectional transmission, acknowledgment segments MAY be interleaved with data segments.¶
The states of a normal TCPCL session (i.e., without session failures) are indicated in Figure 4.¶
+-------+ | START | +-------+ | TCP Establishment | V +-----------+ +---------------------+ | TCP |----------->| Contact / Session | | Connected | | Negotiation | +-----------+ +---------------------+ | +-----Session Parameters-----+ | Negotiated V +-------------+ +-------------+ | Established |----New Transfer---->| Established | | Session | | Session | | Idle |<---Transfers Done---| Live | +-------------+ +-------------+ | | +------------------------------------+ | V +-------------+ | Established | +-------------+ | Session |----Transfers------>| TCP | | Ending | Done | Terminating | +-------------+ +-------------+ | +----------TCP Close Message----------+ | V +-------+ | END | +-------+
Notes on Established Session states:¶
Contact negotiation involves exchanging a Contact Header (CH) in both directions and deriving a negotiated state from the two headers. The contact negotiation sequencing is performed either as the active or passive entity, and is illustrated in Figure 5 and Figure 6 respectively which both share the data validation and negotiation of the Processing of Contact Header "[PCH]" activity of Figure 7 and the "[TCPCLOSE]" activity which indicates TCP connection close. Successful negotiation results in one of the Session Initiation "[SI]" activities being performed. To avoid data loss, a Session Termination "[ST]" exchange allows cleanly finishing transfers before a session is ended.¶
+-------+ | START | +-------+ | TCP Connecting V +-----------+ | TCP | +---------+ | Connected |--Send CH-->| Waiting |--Timeout-->[TCPCLOSE] +-----------+ +---------+ | Received CH V [PCH]
+-----------+ +---------+ | TCP |--Wait for-->| Waiting |--Timeout-->[TCPCLOSE] | Connected | CH +---------+ +-----------+ | Received CH V +-----------------+ | Preparing reply |--Send CH-->[PCH] +-----------------+
+-----------+ | Peer CH | | available | +-----------+ | Validate and Negotiate V +------------+ | Negotiated |--Failure-->[TCPCLOSE] +------------+ | | No TLS +----Negotiate---+ [ST] | TLS | ^ V | Failure +-----------+ V | | TCPCL | +---------------+ | Messaging |<--Success--| TLS Handshake | | Available | +---------------+ +-----------+
Session negotiation involves exchanging a session initialization (SESS_INIT) message in both directions and deriving a negotiated state from the two messages. The session negotiation sequencing is performed either as the active or passive entity, and is illustrated in Figure 8 and Figure 9 respectively which both share the data validation and negotiation of Figure 10. The validation here includes certificate validation and authentication when TLS is used for the session.¶
+-----------+ | TCPCL | +---------+ | Messaging |--Send SESS_INIT-->| Waiting |--Timeout-->[ST] | Available | +---------+ +-----------+ | Received SESS_INIT | V [PSI]
+-----------+ | TCPCL | +---------+ | Messaging |----Wait for ---->| Waiting |--Timeout-->[ST] | Available | SESS_INIT +---------+ +-----------+ | Received SESS_INIT | +-----------------+ | Preparing reply |--Send SESS_INIT-->[PSI] +-----------------+
+----------------+ | Peer SESS_INIT | | available | +----------------+ | Validate and Negotiate V +------------+ | Negotiated |---Failure--->[ST] +------------+ | Success V +--------------+ | Established | | Session Idle | +--------------+
Transfers can occur after a session is established and it's not in the Ending state. Each transfer occurs within a single logical transfer stream between a sender and a receiver, as illustrated in Figure 11 and Figure 12 respectively.¶
+--Send XFER_SEGMENT--+ +--------+ | | | Stream | +-------------+ | | Idle |---Send XFER_SEGMENT-->| In Progress |<------------+ +--------+ +-------------+ | +---------All segments sent-------+ | V +---------+ +--------+ | Waiting |---- Receive Final---->| Stream | | for Ack | XFER_ACK | IDLE | +---------+ +--------+
Notes on transfer sending:¶
+-Receive XFER_SEGMENT-+ +--------+ | Send XFER_ACK | | Stream | +-------------+ | | Idle |--Receive XFER_SEGMENT-->| In Progress |<-------------+ +--------+ +-------------+ | +--------Sent Final XFER_ACK--------+ | V +--------+ | Stream | | Idle | +--------+
Session termination involves one entity initiating the termination of the session and the other entity acknowledging the termination. For either entity, it is the sending of the SESS_TERM message which transitions the session to the Ending substate. While a session is in the Ending state only in-progress transfers can be completed and no new transfers can be started.¶
+-----------+ +---------+ | Session |--Send SESS_TERM-->| Session | | Live/Idle | | Ending | +-----------+ +---------+
+-----------+ +---------+ | Session |--Send SESS_TERM-->| Session | | Live/Idle | | Ending | +-----------+<------+ +---------+ | | Receive SESS_TERM | | | +-------------+
This specification gives requirements about how to use PKIX certificates issued by a Certificate Authority (CA), but does not define any mechanisms for how those certificates come to be. The requirements about TCPCL certificate use are broad to support two quite different PKIX environments:¶
Even within a single TCPCL session, each entity may operate within different PKI environments and with different identifier limitations. The requirements related to identifiers in in a PKIX certificate are in Section 4.4.1.¶
It is important for interoperability that a TCPCL entity have its own security policy tailored to accommodate the peers with which it is expected to operate. A strict TLS security policy is appropriate for a private network with a single shared CA. Operation on the Internet (such as inter-site BP gateways) could trade more lax TCPCL security with the use of encrypted bundle encapsulation [I-D.ietf-dtn-bibect] to ensure strong bundle security.¶
This specification gives requirements about how to initiate, sustain, and terminate a TCPCL session but does not impose any requirements on how sessions need to be managed by a BP agent. It is a network administration matter to determine an appropriate session keeping policy, but guidance given here can be used to steer policy toward performance goals.¶
Many other policies can be established in a TCPCL network between the two extremes of single persistent sessions and only ephemeral sessions. Different policies can be applied to each peer entity and to each bundle as it needs to be transferred (e.g for quality of service). Additionally, future session extension types can apply further nuance to session policies and policy negotiation.¶
Each TCPCL session allows a negotiated transfer segmentation policy to be applied in each transfer direction. A receiving entity can set the Segment MRU in its SESS_INIT message to determine the largest acceptable segment size, and a transmitting entity can segment a transfer into any sizes smaller than the receiver's Segment MRU. It is a network administration matter to determine an appropriate segmentation policy for entities operating TCPCL, but guidance given here can be used to steer policy toward performance goals. It is also advised to consider the Segment MRU in relation to chunking/packetization performed by TLS, TCP, and any intermediate network-layer nodes.¶
Many other policies can be established in a TCPCL network between the two extremes of minimum overhead (large MRU, single-segment) and predictable message sizing (small MRU, highly segmented). Different policies can be applied to each transfer stream to and from any particular entity. Additionally, future session extension and transfer extension types can apply further nuance to transfer policies and policy negotiation.¶
The following figure depicts the protocol exchange for a simple session, showing the session establishment and the transmission of a single bundle split into three data segments (of lengths "L1", "L2", and "L3") from Entity A to Entity B.¶
Note that the sending entity can transmit multiple XFER_SEGMENT messages without waiting for the corresponding XFER_ACK responses. This enables pipelining of messages on a transfer stream. Although this example only demonstrates a single bundle transmission, it is also possible to pipeline multiple XFER_SEGMENT messages for different bundles without necessarily waiting for XFER_ACK messages to be returned for each one. However, interleaving data segments from different bundles is not allowed.¶
No errors or rejections are shown in this example.¶
Entity A Entity B ======== ======== +-------------------------+ | Open TCP Connection | -> +-------------------------+ +-------------------------+ <- | Accept Connection | +-------------------------+ +-------------------------+ | Contact Header | -> +-------------------------+ +-------------------------+ <- | Contact Header | +-------------------------+ +-------------------------+ | SESS_INIT | -> +-------------------------+ +-------------------------+ <- | SESS_INIT | +-------------------------+ +-------------------------+ | XFER_SEGMENT (start) | -> | Transfer ID [I1] | | Length [L1] | | Bundle Data 0..(L1-1) | +-------------------------+ +-------------------------+ +-------------------------+ | XFER_SEGMENT | -> <- | XFER_ACK (start) | | Transfer ID [I1] | | Transfer ID [I1] | | Length [L2] | | Length [L1] | |Bundle Data L1..(L1+L2-1)| +-------------------------+ +-------------------------+ +-------------------------+ +-------------------------+ | XFER_SEGMENT (end) | -> <- | XFER_ACK | | Transfer ID [I1] | | Transfer ID [I1] | | Length [L3] | | Length [L1+L2] | |Bundle Data | +-------------------------+ | (L1+L2)..(L1+L2+L3-1)| +-------------------------+ +-------------------------+ <- | XFER_ACK (end) | | Transfer ID [I1] | | Length [L1+L2+L3] | +-------------------------+ +-------------------------+ | SESS_TERM | -> +-------------------------+ +-------------------------+ <- | SESS_TERM | +-------------------------+ +-------------------------+ +-------------------------+ | TCP Close | -> <- | TCP Close | +-------------------------+ +-------------------------+
For bundle transmissions to occur using the TCPCL, a TCPCL session MUST first be established between communicating entities. It is up to the implementation to decide how and when session setup is triggered. For example, some sessions can be opened proactively and maintained for as long as is possible given the network conditions, while other sessions are be opened only when there is a bundle that is queued for transmission and the routing algorithm selects a certain next-hop node.¶
To establish a TCPCL session, an entity MUST first establish a TCP connection with the intended peer entity, typically by using the services provided by the operating system. Destination port number 4556 has been assigned by IANA as the Registered Port number for the TCP convergence layer. Other destination port numbers MAY be used per local configuration. Determining a peer's destination port number (if different from the registered TCPCL port number) is up to the implementation. Any source port number MAY be used for TCPCL sessions. Typically an operating system assigned number in the TCP Ephemeral range (49152-65535) is used.¶
If the entity is unable to establish a TCP connection for any reason, then it is an implementation matter to determine how to handle the connection failure. An entity MAY decide to re-attempt to establish the connection. If it does so, it MUST NOT overwhelm its target with repeated connection attempts. Therefore, the entity MUST NOT retry the connection setup earlier than some delay time from the last attempt, and it SHOULD use a (binary) exponential back-off mechanism to increase this delay in case of repeated failures. The upper limit on a re-attempt back-off is implementation defined but SHOULD be no longer than one minute (60 seconds) before signaling to the BP agent that a connection cannot be made.¶
Once a TCP connection is established, the active entity SHALL immediately transmit its Contact Header. Once a TCP connection is established, the passive entity SHALL wait for the peer's Contact Header. If the passive entity does not receive a Contact Header after some implementation-defined time duration after TCP connection is established, the entity SHALL close the TCP connection. Entities SHOULD choose a Contact Header reception timeout interval no longer than one minute (60 seconds). Upon reception of a Contact Header, the passive entity SHALL transmit its Contact Header. The ordering of the Contact Header exchange allows the passive entity to avoid allocating resources to a potential TCPCL session until after a valid Contact Header has been received from the active entity. This ordering also allows the passive peer to adapt to alternate TCPCL protocol versions.¶
The format of the Contact Header is described in Section 4.2. Because the TCPCL protocol version in use is part of the initial Contact Header, entities using TCPCL version 4 can coexist on a network with entities using earlier TCPCL versions (with some negotiation needed for interoperation as described in Section 4.3).¶
Within this specification when an entity is said to "close" a TCP connection the entity SHALL use the TCP FIN mechanism and not the RST mechanism. Either mechanism, however, when received will cause a TCP connection to become closed.¶
This section describes the format of the Contact Header and the meaning of its fields.¶
If an entity is capable of exchanging messages according to TLS 1.3 [RFC8446] or any successors which are compatible with that TLS ClientHello, the the CAN_TLS flag within its Contact Header SHALL be set to 1. This behavior prefers the use of TLS when possible, even if security policy does not allow or require authentication. This follows the opportunistic security model of [RFC7435], though an active attacker could interfere with the exchange in such cases.¶
Upon receipt of the Contact Header, both entities perform the validation and negotiation procedures defined in Section 4.3. After receiving the Contact Header from the other entity, either entity MAY refuse the session by sending a SESS_TERM message with an appropriate reason code.¶
The format for the Contact Header is as follows:¶
1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+---------------+---------------+ | magic='dtn!' | +---------------+---------------+---------------+---------------+ | Version | Flags | +---------------+---------------+
See Section 4.3 for details on the use of each of these Contact Header fields.¶
The fields of the Contact Header are:¶
Name | Code | Description |
---|---|---|
CAN_TLS | 0x01 | If bit is set, indicates that the sending peer is capable of TLS security. |
Reserved | others |
Upon reception of the Contact Header, each entity follows the following procedures to ensure the validity of the TCPCL session and to negotiate values for the session parameters.¶
If the magic string is not present or is not valid, the connection MUST be terminated. The intent of the magic string is to provide some protection against an inadvertent TCP connection by a different protocol than the one described in this document. To prevent a flood of repeated connections from a misconfigured application, a passive entity MAY deny new TCP connections from a specific peer address for a period of time after one or more connections fail to provide a decodable Contact Header.¶
The first negotiation is on the TCPCL protocol version to use. The active entity always sends its Contact Header first and waits for a response from the passive entity. During contact initiation, the active TCPCL entity SHALL send the highest TCPCL protocol version on a first session attempt for a TCPCL peer. If the active entity receives a Contact Header with a lower protocol version than the one sent earlier on the TCP connection, the TCP connection SHALL be closed. If the active entity receives a SESS_TERM message with reason of "Version Mismatch", that entity MAY attempt further TCPCL sessions with the peer using earlier protocol version numbers in decreasing order. Managing multi-TCPCL-session state such as this is an implementation matter.¶
If the passive entity receives a Contact Header containing a version that is not a version of the TCPCL that the entity implements, then the entity SHALL send its Contact Header and immediately terminate the session with a reason code of "Version mismatch". If the passive entity receives a Contact Header with a version that is lower than the latest version of the protocol that the entity implements, the entity MAY either terminate the session (with a reason code of "Version mismatch") or adapt its operation to conform to the older version of the protocol. The decision of version fall-back is an implementation matter.¶
The negotiated contact parameters defined by this specification are described in the following paragraphs.¶
This version of the TCPCL supports establishing a Transport Layer Security (TLS) session within an existing TCP connection. When TLS is used within the TCPCL it affects the entire session. Once TLS is established, there is no mechanism available to downgrade the TCPCL session to non-TLS operation.¶
Once established, the lifetime of a TLS connection SHALL be bound to the lifetime of the underlying TCP connection. Immediately prior to actively ending a TLS connection after TCPCL session termination, the peer which sent the original (non-reply) SESS_TERM message SHOULD follow the Closure Alert procedure of [RFC8446] to cleanly terminate the TLS connection. Because each TCPCL message is either fixed-length or self-indicates its length, the lack of a TLS Closure Alert will not cause data truncation or corruption.¶
Subsequent TCPCL session attempts to the same passive entity MAY attempt use the TLS connection resumption feature. There is no guarantee that the passive entity will accept the request to resume a TLS session, and the active entity cannot assume any resumption outcome.¶
The TCPCL uses TLS for certificate exchange in both directions to identify each entity and to allow each entity to authenticate its peer. Each certificate can potentially identify multiple entities and there is no problem using such a certificate as long as the identifiers are sufficient to meet authentication policy (as described in later sections) for the entity which presents it.¶
Because the PKIX environment of each TCPCL entity are likely not controlled by the certificate end users (see Section 3.4), the TCPCL defines a prioritized list of what a certificate can identify about a TCPCL entity:¶
When only a DNS-ID or IPADDR-ID can be identified by a certificate, it is implied that an entity which authenticates using that certificate is trusted to provide a valid Node ID in its SESS_INIT; the certificate itself does not actually authenticate that Node ID.¶
The RECOMMENDED security policy of an entity is to validate a NODE-ID when present, and only require DNS-ID/IPADDR-ID authentication in the absence of a NODE-ID.¶
This specification defines a NODE-ID of a certificate as being the subjectAltName entry of type uniformResourceIdentifier whose value is a URI consistent with the requirements of [RFC3986] and the URI schemes of the IANA "Bundle Protocol URI Scheme Type" registry [IANA-BUNDLE]. This is similar to the URI-ID of [RFC6125] but does not require any structure to the scheme-specific-part of the URI. Unless specified otherwise by the definition of the URI scheme being authenticated, URI matching of a NODE-ID SHALL use the URI comparison logic of [RFC3986] and scheme-based normalization of those schemes specified in [I-D.ietf-dtn-bpbis]. A URI scheme can refine this "exact match" logic with rules about how Node IDs within that scheme are to be compared with the certificate-authenticated NODE-ID.¶
This specification defines a IPADDR-ID of a certificate as being the subjectAltName entry of type iPAddress whose value is encoded according to [RFC5280].¶
The use of TLS is negotiated using the Contact Header as described in Section 4.3. After negotiating an Enable TLS parameter of true, and before any other TCPCL messages are sent within the session, the session entities SHALL begin a TLS handshake in accordance with [RFC8446]. By convention, this protocol uses the entity which initiated the underlying TCP connection (the active peer) as the "client" role of the TLS handshake request.¶
The TLS handshake, if it occurs, is considered to be part of the contact negotiation before the TCPCL session itself is established. Specifics about sensitive data exposure are discussed in Section 8.¶
The parameters within each TLS negotiation are implementation dependent but any TCPCL entity SHALL follow all recommended practices of BCP 195 [RFC7525], or any updates or successors that become part of BCP 195. Within each TLS handshake, the following requirements apply (using the rough order in which they occur):¶
All certificates supplied during TLS handshake SHALL conform to [RFC5280], or any updates or successors to that profile. When a certificate is supplied during TLS handshake, the full certification chain SHOULD be included unless security policy indicates that is unnecessary.¶
If a TLS handshake cannot negotiate a TLS connection, both entities of the TCPCL session SHALL close the TCP connection. At this point the TCPCL session has not yet been established so there is no TCPCL session to terminate.¶
After a TLS connection is successfully established, the active entity SHALL send a SESS_INIT message to begin session negotiation. This session negotiation and all subsequent messaging are secured.¶
Using PKIX certificates exchanged during the TLS handshake, each of the entities can authenticate a peer Node ID directly or authenticate the peer DNS name or network address. The logic for handling certificates and certificate data is separated into three phases:¶
By using the SNI DNS name (see Section 4.4.2) a single passive entity can act as a convergence layer for multiple BP agents with distinct Node IDs. When this "virtual host" behavior is used, the DNS name is used as the indication of which BP Node the active entity is attempting to communicate with. A virtual host CL entity can be authenticated by a certificate containing all of the DNS names and/or Node IDs being hosted or by several certificates each authenticating a single DNS name and/or Node ID, using the SNI value from the peer to select which certificate to use.¶
For any peer certificate received during TLS handshake, the entity SHALL perform the certification path validation of [RFC5280] up to one of the entity's trusted CA certificates. If certificate validation fails or if security policy disallows a certificate for any reason, the entity SHALL fail the TLS handshake with a bad certificate error. Leaving out part of the certification chain can cause the entity to fail to validate a certificate if the left-out certificates are unknown to the entity (see Section 8.6).¶
The result of authenticating a peer identity (i.e., Node ID, DNS name, or IP address) against one type of certificate claim is one of:¶
Either during or immediately after the TLS handshake if the active entity resolved a DNS name (of the passive entity) in order to initiate the TCP connection, the active entity SHALL authenticate that DNS name using any DNS-ID of the peer certificate. If the DNS name authentication result is Failure or if the result is Absent and security policy requires an authenticated DNS name, the entity SHALL terminate the session (with a reason code of "Contact Failure").¶
Either during or immediately after the TLS handshake, each side of the session SHALL authenticate the IP address of the other side of the TCP connection using any IPADDR-ID of the peer certificate. If the address authentication result is Failure or if the result is Absent and security policy requires an authenticated network address, the entity SHALL terminate the session (with a reason code of "Contact Failure").¶
Immediately before Session Parameter Negotiation, each side of the session SHALL authenticate the Node ID of the peer's SESS_INIT message using any NODE-ID of the peer certificate. If the Node ID authentication result is Failure or if the result is Absent and security policy requires an authenticated Node ID, the entity SHALL terminate the session (with a reason code of "Contact Failure").¶
A summary of a typical TLS use is shown in the sequence in Figure 17 below. In this example the active peer terminates the session but termination can be initiated from either peer.¶
Entity A Entity B active peer passive peer +-------------------------+ | Open TCP Connection | -> +-------------------------+ +-------------------------+ <- | Accept Connection | +-------------------------+ +-------------------------+ | Contact Header | -> +-------------------------+ +-------------------------+ <- | Contact Header | +-------------------------+ +-------------------------+ +-------------------------+ | TLS Negotiation | -> <- | TLS Negotiation | | (as client) | | (as server) | +-------------------------+ +-------------------------+ DNS-ID and IPADDR-ID authentication occurs. Secured TCPCL messaging can begin. +-------------------------+ | SESS_INIT | -> +-------------------------+ +-------------------------+ <- | SESS_INIT | +-------------------------+ NODE-ID authentication occurs. Session is established, transfers can begin. +-------------------------+ | SESS_TERM | -> +-------------------------+ +-------------------------+ <- | SESS_TERM | +-------------------------+ +-------------------------+ | TLS Closure Alert | -> +-------------------------+ +-------------------------+ <- | TLS Closure Alert | +-------------------------+ +-------------------------+ +-------------------------+ | TCP Close | -> <- | TCP Close | +-------------------------+ +-------------------------+
After the initial exchange of a Contact Header and (if TLS is negotiated to be used) the TLS handshake, all messages transmitted over the session are identified by a one-octet header with the following structure:¶
0 1 2 3 4 5 6 7 +---------------+ | Message Type | +---------------+
The message header fields are as follows:¶
Name | Code | Description |
---|---|---|
SESS_INIT | 0x07 | Contains the session parameter inputs from one of the entities, as described in Section 4.6. |
SESS_TERM | 0x05 | Indicates that one of the entities participating in the session wishes to cleanly terminate the session, as described in Section 6.1. |
XFER_SEGMENT | 0x01 | Indicates the transmission of a segment of bundle data, as described in Section 5.2.2. |
XFER_ACK | 0x02 | Acknowledges reception of a data segment, as described in Section 5.2.3. |
XFER_REFUSE | 0x03 | Indicates that the transmission of the current bundle SHALL be stopped, as described in Section 5.2.4. |
KEEPALIVE | 0x04 | Used to keep TCPCL session active, as described in Section 5.1.1. |
MSG_REJECT | 0x06 | Contains a TCPCL message rejection, as described in Section 5.1.2. |
Before a session is established and ready to transfer bundles, the session parameters are negotiated between the connected entities. The SESS_INIT message is used to convey the per-entity parameters which are used together to negotiate the per-session parameters as described in Section 4.7.¶
The format of a SESS_INIT message is as follows in Figure 19.¶
+-----------------------------+ | Message Header | +-----------------------------+ | Keepalive Interval (U16) | +-----------------------------+ | Segment MRU (U64) | +-----------------------------+ | Transfer MRU (U64) | +-----------------------------+ | Node ID Length (U16) | +-----------------------------+ | Node ID Data (variable) | +-----------------------------+ | Session Extension | | Items Length (U32) | +-----------------------------+ | Session Extension | | Items (var.) | +-----------------------------+
The fields of the SESS_INIT message are:¶
When the active entity initiates a TCPCL session, it is likely based on routing information which binds a Node ID to CL parameters. If the active entity receives a SESS_INIT with different Node ID than was intended for the TCPCL session, the session MAY be allowed to be established. If allowed, such a session SHALL be associated with the Node ID provided in the SESS_INIT message rather than any intended value.¶
An entity calculates the parameters for a TCPCL session by negotiating the values from its own preferences (conveyed by the SESS_INIT it sent to the peer) with the preferences of the peer entity (expressed in the SESS_INIT that it received from the peer). The negotiated parameters defined by this specification are described in the following paragraphs.¶
Once this process of parameter negotiation is completed, this protocol defines no additional mechanism to change the parameters of an established session; to effect such a change, the TCPCL session MUST be terminated and a new session established.¶
Each of the Session Extension Items SHALL be encoded in an identical Type-Length-Value (TLV) container form as indicated in Figure 20.¶
The fields of the Session Extension Item are:¶
1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+---------------+---------------+ | Item Flags | Item Type | Item Length...| +---------------+---------------+---------------+---------------+ | length contd. | Item Value... | +---------------+---------------+---------------+---------------+
Name | Code | Description |
---|---|---|
CRITICAL | 0x01 | If bit is set, indicates that the receiving peer must handle the extension item. |
Reserved | others |
This section describes the protocol operation for the duration of an established session, including the mechanism for transmitting bundles over the session.¶
The protocol includes a provision for transmission of KEEPALIVE messages over the TCPCL session to help determine if the underlying TCP connection has been disrupted.¶
As described in Section 4.3, a negotiated parameter of each session is the Session Keepalive interval. If the negotiated Session Keepalive is zero (i.e., one or both contact headers contains a zero Keepalive Interval), then the keepalive feature is disabled. There is no logical minimum value for the keepalive interval (within the minimum imposed by the positive-value encoding), but when used for many sessions on an open, shared network a short interval could lead to excessive traffic. For shared network use, entities SHOULD choose a keepalive interval no shorter than 30 seconds. There is no logical maximum value for the keepalive interval (within the maximum imposed by the fixed-size encoding), but an idle TCP connection is liable for closure by the host operating system if the keepalive time is longer than tens-of-minutes. Entities SHOULD choose a keepalive interval no longer than 10 minutes (600 seconds).¶
Note: The Keepalive Interval SHOULD NOT be chosen too short as TCP retransmissions MAY occur in case of packet loss. Those will have to be triggered by a timeout (TCP retransmission timeout (RTO)), which is dependent on the measured RTT for the TCP connection so that KEEPALIVE messages can experience noticeable latency.¶
The format of a KEEPALIVE message is a one-octet message type code of KEEPALIVE (as described in Table 2) with no additional data. Both sides SHALL send a KEEPALIVE message whenever the negotiated interval has elapsed with no transmission of any message (KEEPALIVE or other).¶
If no message (KEEPALIVE or other) has been received in a session after some implementation-defined time duration, then the entity SHALL terminate the session by transmitting a SESS_TERM message (as described in Section 6.1) with reason code "Idle Timeout". If configurable, the idle timeout duration SHOULD be no shorter than twice the keepalive interval. If not configurable, the idle timeout duration SHOULD be exactly twice the keepalive interval.¶
This message type is not expected to be seen in a well-functioning session. Its purpose is to aid in troubleshooting bad entity behavior by allowing the peer to observe why an entity is not responding as expected to its messages.¶
If a TCPCL entity receives a message type which is unknown to it (possibly due to an unhandled protocol version mismatch or a incorrectly-negotiated session extension which defines a new message type), the entity SHALL send a MSG_REJECT message with a Reason Code of "Message Type Unknown" and close the TCP connection. If a TCPCL entity receives a message type which is known but is inappropriate for the negotiated session parameters (possibly due to incorrectly-negotiated session extension), the entity SHALL send a MSG_REJECT message with a Reason Code of "Message Unsupported". If a TCPCL entity receives a message which is inappropriate for the current session state (e.g., a SESS_INIT after the session has already been established or an XFER_ACK message with an unknown Transfer ID), the entity SHALL send a MSG_REJECT message with a Reason Code of "Message Unexpected".¶
The format of a MSG_REJECT message is as follows in Figure 21.¶
+-----------------------------+ | Message Header | +-----------------------------+ | Reason Code (U8) | +-----------------------------+ | Rejected Message Header | +-----------------------------+
The fields of the MSG_REJECT message are:¶
Name | Code | Description |
---|---|---|
Message Type Unknown | 0x01 | A message was received with a Message Type code unknown to the TCPCL entity. |
Message Unsupported | 0x02 | A message was received but the TCPCL entity cannot comply with the message contents. |
Message Unexpected | 0x03 | A message was received while the session is in a state in which the message is not expected. |
All of the messages in this section are directly associated with transferring a bundle between TCPCL entities.¶
A single TCPCL transfer results in a bundle (handled by the convergence layer as opaque data) being exchanged from one entity to the other. In TCPCL a transfer is accomplished by dividing a single bundle up into "segments" based on the receiving-side Segment MRU (see Section 4.2). The choice of the length to use for segments is an implementation matter, but each segment MUST NOT be larger than the receiving entity's maximum receive unit (MRU) (see the field Segment MRU of Section 4.2). The first segment for a bundle is indicated by the 'START' flag and the last segment is indicated by the 'END' flag.¶
A single transfer (and by extension a single segment) SHALL NOT contain data of more than a single bundle. This requirement is imposed on the agent using the TCPCL rather than TCPCL itself.¶
If multiple bundles are transmitted on a single TCPCL connection, they MUST be transmitted consecutively without interleaving of segments from multiple bundles.¶
Each of the bundle transfer messages contains a Transfer ID which is used to correlate messages (from both sides of a transfer) for each bundle. A Transfer ID does not attempt to address uniqueness of the bundle data itself and has no relation to concepts such as bundle fragmentation. Each invocation of TCPCL by the bundle protocol agent, requesting transmission of a bundle (fragmentary or otherwise), results in the initiation of a single TCPCL transfer. Each transfer entails the sending of a sequence of some number of XFER_SEGMENT and XFER_ACK messages; all are correlated by the same Transfer ID. The sending entity originates a transfer ID and the receiving entity uses that same Transfer ID in acknowledgements.¶
Transfer IDs from each entity SHALL be unique within a single TCPCL session. Upon exhaustion of the entire 64-bit Transfer ID space, the sending entity SHALL terminate the session with SESS_TERM reason code "Resource Exhaustion". For bidirectional bundle transfers, a TCPCL entity SHOULD NOT rely on any relation between Transfer IDs originating from each side of the TCPCL session.¶
Although there is not a strict requirement for Transfer ID initial values or ordering (see Section 8.12), in the absence of any other mechanism for generating Transfer IDs an entity SHALL use the following algorithm: The initial Transfer ID from each entity is zero and subsequent Transfer ID values are incremented from the prior Transfer ID value by one.¶
Each bundle is transmitted in one or more data segments. The format of a XFER_SEGMENT message follows in Figure 22.¶
+------------------------------+ | Message Header | +------------------------------+ | Message Flags (U8) | +------------------------------+ | Transfer ID (U64) | +------------------------------+ | Transfer Extension | | Items Length (U32) | | (only for START segment) | +------------------------------+ | Transfer Extension | | Items (var.) | | (only for START segment) | +------------------------------+ | Data length (U64) | +------------------------------+ | Data contents (octet string) | +------------------------------+
The fields of the XFER_SEGMENT message are:¶
Name | Code | Description |
---|---|---|
END | 0x01 | If bit is set, indicates that this is the last segment of the transfer. |
START | 0x02 | If bit is set, indicates that this is the first segment of the transfer. |
Reserved | others |
The flags portion of the message contains two flag values in the two low-order bits, denoted 'START' and 'END' in Table 5. The 'START' flag SHALL be set to 1 when transmitting the first segment of a transfer. The 'END' flag SHALL be set to 1 when transmitting the last segment of a transfer. In the case where an entire transfer is accomplished in a single segment, both the 'START' and 'END' flags SHALL be set to 1.¶
Once a transfer of a bundle has commenced, the entity MUST only send segments containing sequential portions of that bundle until it sends a segment with the 'END' flag set to 1. No interleaving of multiple transfers from the same entity is possible within a single TCPCL session. Simultaneous transfers between two entities MAY be achieved using multiple TCPCL sessions.¶
Although the TCP transport provides reliable transfer of data between transport peers, the typical BSD sockets interface provides no means to inform a sending application of when the receiving application has processed some amount of transmitted data. Thus, after transmitting some data, the TCPCL needs an additional mechanism to determine whether the receiving agent has successfully received and fully processed the segment. To this end, the TCPCL protocol provides feedback messaging whereby a receiving entity transmits acknowledgments of reception of data segments.¶
The format of an XFER_ACK message follows in Figure 23.¶
+-----------------------------+ | Message Header | +-----------------------------+ | Message Flags (U8) | +-----------------------------+ | Transfer ID (U64) | +-----------------------------+ | Acknowledged length (U64) | +-----------------------------+
The fields of the XFER_ACK message are:¶
A receiving TCPCL entity SHALL send an XFER_ACK message in response to each received XFER_SEGMENT message after the segment has been fully processed. The flags portion of the XFER_ACK header SHALL be set to match the corresponding XFER_SEGMENT message being acknowledged (including flags not decodable to the entity). The acknowledged length of each XFER_ACK contains the sum of the data length fields of all XFER_SEGMENT messages received so far in the course of the indicated transfer. The sending entity SHOULD transmit multiple XFER_SEGMENT messages without waiting for the corresponding XFER_ACK responses. This enables pipelining of messages on a transfer stream.¶
For example, suppose the sending entity transmits four segments of bundle data with lengths 100, 200, 500, and 1000, respectively. After receiving the first segment, the entity sends an acknowledgment of length 100. After the second segment is received, the entity sends an acknowledgment of length 300. The third and fourth acknowledgments are of length 800 and 1800, respectively.¶
The TCPCL supports a mechanism by which a receiving entity can indicate to the sender that it does not want to receive the corresponding bundle. To do so, upon receiving an XFER_SEGMENT message, the entity MAY transmit a XFER_REFUSE message. As data segments and acknowledgments can cross on the wire, the bundle that is being refused SHALL be identified by the Transfer ID of the refusal.¶
There is no required relation between the Transfer MRU of a TCPCL entity (which is supposed to represent a firm limitation of what the entity will accept) and sending of a XFER_REFUSE message. A XFER_REFUSE can be used in cases where the agent's bundle storage is temporarily depleted or somehow constrained. A XFER_REFUSE can also be used after the bundle header or any bundle data is inspected by an agent and determined to be unacceptable.¶
A transfer receiver MAY send an XFER_REFUSE message as soon as it receives any XFER_SEGMENT message. The transfer sender MUST be prepared for this and MUST associate the refusal with the correct bundle via the Transfer ID fields.¶
The TCPCL itself does not have any required behavior to respond to an XFER_REFUSE based on its Reason Code; the refusal is passed up as an indication to the BP agent that the transfer has been refused. If a transfer refusal has a Reason Code which is not decodable to the BP agent, the agent SHOULD treat the refusal as having an Unknown reason.¶
The format of the XFER_REFUSE message is as follows in Figure 24.¶
+-----------------------------+ | Message Header | +-----------------------------+ | Reason Code (U8) | +-----------------------------+ | Transfer ID (U64) | +-----------------------------+
The fields of the XFER_REFUSE message are:¶
Name | Code | Description |
---|---|---|
Unknown | 0x00 | Reason for refusal is unknown or not specified. |
Completed | 0x01 | The receiver already has the complete bundle. The sender MAY consider the bundle as completely received. |
No Resources | 0x02 | The receiver's resources are exhausted. The sender SHOULD apply reactive bundle fragmentation before retrying. |
Retransmit | 0x03 | The receiver has encountered a problem that requires the bundle to be retransmitted in its entirety. |
Not Acceptable | 0x04 | Some issue with the bundle data or the transfer extension data was encountered. The sender SHOULD NOT retry the same bundle with the same extensions. |
Extension Failure | 0x05 | A failure processing the Transfer Extension Items has occurred. |
Session Terminating | 0x06 | The receiving entity is in the process of terminating the session. The sender MAY retry the same bundle at a later time in a different session. |
The receiver MUST, for each transfer preceding the one to be refused, have either acknowledged all XFER_SEGMENT messages or refused the bundle transfer.¶
The bundle transfer refusal MAY be sent before an entire data segment is received. If a sender receives a XFER_REFUSE message, the sender MUST complete the transmission of any partially sent XFER_SEGMENT message. There is no way to interrupt an individual TCPCL message partway through sending it. The sender MUST NOT commence transmission of any further segments of the refused bundle subsequently. Note, however, that this requirement does not ensure that an entity will not receive another XFER_SEGMENT for the same bundle after transmitting a XFER_REFUSE message since messages can cross on the wire; if this happens, subsequent segments of the bundle SHALL also be refused with a XFER_REFUSE message.¶
Note: If a bundle transmission is aborted in this way, the receiver does not receive a segment with the 'END' flag set to 1 for the aborted bundle. The beginning of the next bundle is identified by the 'START' flag set to 1, indicating the start of a new transfer, and with a distinct Transfer ID value.¶
Each of the Transfer Extension Items SHALL be encoded in an identical Type-Length-Value (TLV) container form as indicated in Figure 25.¶
The fields of the Transfer Extension Item are:¶
1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+---------------+---------------+ | Item Flags | Item Type | Item Length...| +---------------+---------------+---------------+---------------+ | length contd. | Item Value... | +---------------+---------------+---------------+---------------+
Name | Code | Description |
---|---|---|
CRITICAL | 0x01 | If bit is set, indicates that the receiving peer must handle the extension item. |
Reserved | others |
The purpose of the Transfer Length extension is to allow entities to preemptively refuse bundles that would exceed their resources or to prepare storage on the receiving entity for the upcoming bundle data.¶
Multiple Transfer Length extension items SHALL NOT occur within the same transfer. The lack of a Transfer Length extension item in any transfer SHALL NOT imply anything about the potential length of the transfer. The Transfer Length extension SHALL be assigned transfer extension type ID 0x0001.¶
If a transfer occupies exactly one segment (i.e., both START and END flags are 1) the Transfer Length extension SHOULD NOT be present. The extension does not provide any additional information for single-segment transfers.¶
The format of the Transfer Length data is as follows in Figure 26.¶
+----------------------+ | Total Length (U64) | +----------------------+
The fields of the Transfer Length extension are:¶
This section describes the procedures for terminating a TCPCL session. The purpose of terminating a session is to allow transfers to complete before the TCP connection is closed but not allow any new transfers to start. A session state change is necessary for this to happen because transfers can be in-progress in either direction (transfer stream) within a session. Waiting for a transfer to complete in one direction does not control or influence the possibility of a transfer in the other direction. Either peer of a session can terminate an established session at any time.¶
To cleanly terminate a session, a SESS_TERM message SHALL be transmitted by either entity at any point following complete transmission of any other message. When sent to initiate a termination, the REPLY flag of a SESS_TERM message SHALL be 0. Upon receiving a SESS_TERM message after not sending a SESS_TERM message in the same session, an entity SHALL send an acknowledging SESS_TERM message. When sent to acknowledge a termination, a SESS_TERM message SHALL have identical data content from the message being acknowledged except for the REPLY flag, which is set to 1 to indicate acknowledgement.¶
Once a SESS_TERM message is sent the state of that TCPCL session changes to Ending. While the session is in the Ending state, an entity MAY finish an in-progress transfer in either direction. While the session is in the Ending state, an entity SHALL NOT begin any new outgoing transfer for the remainder of the session. While the session is in the Ending state, an entity SHALL NOT accept any new incoming transfer for the remainder of the session. If a new incoming transfer is attempted while in the Ending state, the receiving entity SHALL send an XFER_REFUSE with a Reason Code of "Session Terminating".¶
There are circumstances where an entity has an urgent need to close a TCP connection associated with a TCPCL session, without waiting for transfers to complete but also in a way which doesn't force timeouts to occur; for example, due to impending shutdown of the underlying data link layer. Instead of following a clean termination sequence, after transmitting a SESS_TERM message an entity MAY perform an unclean termination by immediately closing the associated TCP connection. When performing an unclean termination, an entity SHOULD acknowledge all received XFER_SEGMENTs with an XFER_ACK before closing the TCP connection. Not acknowledging received segments can result in unnecessary bundle or bundle fragment retransmission. Any delay between request to close the TCP connection and actual closing of the connection (a "half-closed" state) MAY be ignored by the TCPCL entity. If the underlying TCP connection is closed during a transmission (in either transfer stream), the transfer SHALL be indicated to the BP agent as failed (see the transmission failure and reception failure indications of Section 3.1).¶
The TCPCL itself does not have any required behavior to respond to an SESS_TERM based on its Reason Code; the termination is passed up as an indication to the BP agent that the session state has changed. If a termination has a Reason Code which is not decodable to the BP agent, the agent SHOULD treat the termination as having an Unknown reason.¶
The format of the SESS_TERM message is as follows in Figure 27.¶
+-----------------------------+ | Message Header | +-----------------------------+ | Message Flags (U8) | +-----------------------------+ | Reason Code (U8) | +-----------------------------+
The fields of the SESS_TERM message are:¶
Name | Code | Description |
---|---|---|
REPLY | 0x01 | If bit is set, indicates that this message is an acknowledgement of an earlier SESS_TERM message. |
Reserved | others |
Name | Code | Description |
---|---|---|
Unknown | 0x00 | A termination reason is not available. |
Idle timeout | 0x01 | The session is being terminated due to idleness. |
Version mismatch | 0x02 | The entity cannot conform to the specified TCPCL protocol version. |
Busy | 0x03 | The entity is too busy to handle the current session. |
Contact Failure | 0x04 | The entity cannot interpret or negotiate a Contact Header or SESS_INIT option. |
Resource Exhaustion | 0x05 | The entity has run into some resource limit and cannot continue the session. |
The earliest a TCPCL session termination MAY occur is immediately after transmission of a Contact Header (and prior to any further message transmit). This can, for example, be used to notify that the entity is currently not able or willing to communicate. However, an entity MUST always send the Contact Header to its peer before sending a SESS_TERM message.¶
Termination of the TCP connection MAY occur prior to receiving the Contact header as discussed in Section 4.1. If reception of the Contact Header itself somehow fails (e.g., an invalid "magic string" is received), an entity SHALL close the TCP connection without sending a SESS_TERM message.¶
If a session is to be terminated before a protocol message has completed being sent, then the entity MUST NOT transmit the SESS_TERM message but still SHALL close the TCP connection. Each TCPCL message is contiguous in the octet stream and has no ability to be cut short and/or preempted by an other message. This is particularly important when large segment sizes are being transmitted; either entire XFER_SEGMENT is sent before a SESS_TERM message or the connection is simply terminated mid-XFER_SEGMENT.¶
The protocol includes a provision for clean termination of idle sessions. Determining the length of time to wait before terminating idle sessions, if they are to be terminated at all, is an implementation and configuration matter.¶
If there is a configured time to terminate idle sessions and if no TCPCL messages (other than KEEPALIVE messages) has been received for at least that amount of time, then either entity MAY terminate the session by transmitting a SESS_TERM message indicating the reason code of "Idle timeout" (as described in Table 9).¶
[NOTE to the RFC Editor: please remove this section before publication, as well as the reference to [RFC7942] and [github-dtn-bpbis-tcpcl].]¶
This section records the status of known implementations of the protocol defined by this specification at the time of posting of this Internet-Draft, and is based on a proposal described in [RFC7942]. The description of implementations in this section is intended to assist the IETF in its decision processes in progressing drafts to RFCs. Please note that the listing of any individual implementation here does not imply endorsement by the IETF. Furthermore, no effort has been spent to verify the information presented here that was supplied by IETF contributors. This is not intended as, and must not be construed to be, a catalog of available implementations or their features. Readers are advised to note that other implementations can exist.¶
An example implementation of the this draft of TCPCLv4 has been created as a GitHub project [github-dtn-bpbis-tcpcl] and is intended to use as a proof-of-concept and as a possible source of interoperability testing. This example implementation uses D-Bus as the CL-BP Agent interface, so it only runs on hosts which provide the Python "dbus" library.¶
This section separates security considerations into threat categories based on guidance of BCP 72 [RFC3552].¶
When used without TLS security, the TCPCL exposes the Node ID and other configuration data to passive eavesdroppers. This occurs even when no transfers occur within a TCPCL session. This can be avoided by always using TLS, even if authentication is not available (see Section 8.11).¶
TCPCL can be used to provide point-to-point transport security, but does not provide security of data-at-rest and does not guarantee end-to-end bundle security. The bundle security mechanisms defined in [I-D.ietf-dtn-bpsec] are to be used instead.¶
When used without TLS security, the TCPCL exposes all bundle data to passive eavesdroppers. This can be avoided by always using TLS, even if authentication is not available (see Section 8.11).¶
When a TCPCL entity supports multiple versions of the protocol it is possible for a malicious or misconfigured peer to use an older version of TCPCL which does not support transport security. A man-in-the-middle attacker can also manipulate a Contact Header to present a lower protocol version than desired.¶
It is up to security policies within each TCPCL entity to ensure that the negotiated TCPCL version meets transport security requirements.¶
When security policy allows non-TLS sessions, TCPCL does not protect against active network attackers. It is possible for a man-in-the-middle attacker to set the CAN_TLS flag to 0 on either side of the Contact Header exchange. This leads to the "SSL Stripping" attack described in [RFC7457].¶
The purpose of the CAN_TLS flag is to allow the use of TCPCL on entities which simply do not have a TLS implementation available. When TLS is available on an entity, it is strongly encouraged that the security policy disallow non-TLS sessions. This requires that the TLS handshake occurs, regardless of the policy-driven parameters of the handshake and policy-driven handling of the handshake outcome.¶
One mechanism to mitigate the possibility of TLS stripping is the use of DNS-based Authentication of Named Entities (DANE) [RFC6698] for the passive peer. This mechanism relies on DNS and is unidirectional, so it doesn't help with applying policy toward the active peer, but it can be useful in an environment using opportunistic security. The configuration and use of DANE are are outside of the scope of this document.¶
The negotiated use of TLS is identical behavior to STARTTLS use in [RFC2595] and [RFC4511].¶
Even when using TLS to secure the TCPCL session, the actual ciphersuite negotiated between the TLS peers can be insecure. Recommendations for ciphersuite use are included in BCP 195 [RFC7525]. It is up to security policies within each TCPCL entity to ensure that the negotiated TLS ciphersuite meets transport security requirements.¶
The profile in Section 4.4.3 uses end-entity certificates chained up to a trusted root CA. During TLS handshake, either entity can send a certificate set which does not contain the full chain, possibly excluding intermediate or root CAs. In an environment where peers are known to already contain needed root and intermediate CAs there is no need to include those CAs, but this has a risk of an entity not actually having one of the needed CAs.¶
Even when TLS itself is operating properly an attacker can attempt to exploit vulnerabilities within certificate check algorithms or configuration to establish a secure TCPCL session using an invalid certificate. A BP agent treats the peer Node ID within a TCPCL session as authoritative and an invalid certificate exploit could lead to bundle data leaking and/or denial of service to the Node ID being impersonated. There are many reasons, described in [RFC5280], why a certificate can fail to validate, including using the certificate outside of its valid time interval, using purposes for which it was not authorized, or using it after it has been revoked by its CA. Validating a certificate is a complex task and can require network connectivity outside of the primary TCPCL network path(s) if a mechanism such as the Online Certificate Status Protocol (OCSP) [RFC6960] is used by the CA. The configuration and use of particular certificate validation methods are outside of the scope of this document.¶
Even with a secure block cipher and securely-established session keys, there are limits to the amount of plaintext which can be safely encrypted with a given set of keys as described in [AEAD-LIMITS]. When permitted by the negotiated TLS version (see [RFC8446]), it is advisable to take advantage of session key updates to avoid those limits.¶
The certificates exchanged by TLS enable authentication of peer DNS name and Node ID, but it is possible that a peer either not provide a valid certificate or that the certificate does not validate either the DNS name or Node ID of the peer (see Section 3.4). Having a CA-validated certificate does not alone guarantee the identity of the network host or BP node from which the certificate is provided; additional validation procedures in Section 4.4.2 bind the DNS name or Node ID based on the contents of the certificate.¶
The DNS name validation is a weaker form of authentication, because even if a peer is operating on an authenticated network DNS name it can provide an invalid Node ID and cause bundles to be "leaked" to an invalid node. Especially in DTN environments, network names and addresses of nodes can be time-variable so binding a certificate to a Node ID is a more stable identity.¶
Node ID validation ensures that the peer to which a bundle is transferred is in fact the node which the BP Agent expects it to be. It is a reasonable policy to skip DNS name validation if certificates can be guaranteed to validate the peer's Node ID. In circumstances where certificates can only be issued to DNS names, Node ID validation is not possible but it could be reasonable to assume that a trusted host is not going to present an invalid Node ID. Determining of when a DNS name authentication can be trusted to validate a Node ID is also a policy matter outside the scope of this document.¶
The behaviors described in this section all amount to a potential denial-of-service to a TCPCL entity. The denial-of-service could be limited to an individual TCPCL session, could affect other well-behaving sessions on an entity, or could affect all sessions on a host.¶
A malicious entity can continually establish TCPCL sessions and delay sending of protocol-required data to trigger timeouts. The victim entity can block TCP connections from network peers which are thought to be incorrectly behaving within TCPCL.¶
An entity can send a large amount of data over a TCPCL session, requiring the receiving entity to handle the data. The victim entity can attempt to stop the flood of data by sending an XFER_REFUSE message, or forcibly terminate the session.¶
There is the possibility of a "data dribble" attack in which an entity presents a very small Segment MRU which causes transfers to be split among an large number of very small segments and causes the segmentation overhead to overwhelm the actual bundle data segments. Similarly, an entity can present a very small Transfer MRU which will cause resources to be wasted on establishment and upkeep of a TCPCL session over which a bundle could never be transferred. The victim entity can terminate the session during the negotiation of Section 4.7 if the MRUs are unacceptable.¶
The keepalive mechanism can be abused to waste throughput within a network link which would otherwise be usable for bundle transmissions. Due to the quantization of the Keepalive Interval parameter the smallest Session Keepalive is one second, which should be long enough to not flood the link. The victim entity can terminate the session during the negotiation of Section 4.7 if the Keepalive Interval is unacceptable.¶
Finally, an attacker or a misconfigured entity can cause issues at the TCP connection which will cause unnecessary TCP retransmissions or connection resets, effectively denying the use of the overlying TCPCL session.¶
This specification makes use of PKIX certificate validation and authentication within TLS. There are alternate uses of TLS which are not necessarily incompatible with the security goals of this specification, but are outside of the scope of this document. The following subsections give examples of alternate TLS uses.¶
In environments where PKI is available but there are restrictions on the issuance of certificates (including the contents of certificates), it may be possible to make use of TLS in a way which authenticates only the passive entity of a TCPCL session or which does not authenticate either entity. Using TLS in a way which does not successfully authenticate some claim of both peer entities of a TCPCL session is outside of the scope of this document but does have similar properties to the opportunistic security model of [RFC7435].¶
In environments where PKI is unavailable, alternate uses of TLS which do not require certificates such as pre-shared key (PSK) authentication [RFC5489] and the use of raw public keys [RFC7250] are available and can be used to ensure confidentiality within TCPCL. Using non-PKI node authentication methods is outside of the scope of this document.¶
The only requirement on Transfer IDs is that they be unique with each session from the sending peer only. The trivial algorithm of the first transfer starting at zero and later transfers incrementing by one causes absolutely predictable Transfer IDs. Even when a TCPCL session is not TLS secured and there is a man-in-the-middle attacker causing denial of service with XFER_REFUSE messages, it is not possible to preemptively refuse a transfer so there is no benefit in having unpredictable Transfer IDs within a session.¶
Registration procedures referred to in this section are defined in [RFC8126].¶
Some of the registries have been defined as version specific to TCPCLv4, and imports some or all codepoints from TCPCLv3. This was done to disambiguate the use of these codepoints between TCPCLv3 and TCPCLv4 while preserving the semantics of some of the codepoints.¶
Within the port registry of [IANA-PORTS], TCP port number 4556 has been previously assigned as the default port for the TCP convergence layer in [RFC7242]. This assignment is unchanged by TCPCL version 4, but the assignment reference is updated to this specification. Each TCPCL entity identifies its TCPCL protocol version in its initial contact (see Section 9.2), so there is no ambiguity about what protocol is being used. The related assignments for UDP and DCCP port 4556 (both registered by [RFC7122]) are unchanged.¶
Parameter | Value |
---|---|
Service Name: | dtn-bundle |
Transport Protocol(s): | TCP |
Assignee: | IESG <iesg@ietf.org> |
Contact: | IESG <iesg@ietf.org> |
Description: | DTN Bundle TCP CL Protocol |
Reference: | This specification. |
Port Number: | 4556 |
IANA has created, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version Numbers". The version number table is updated to include this specification. The registration procedure is RFC Required.¶
Value | Description | Reference |
---|---|---|
0 | Reserved | [RFC7242] |
1 | Reserved | [RFC7242] |
2 | Reserved | [RFC7242] |
3 | TCPCL | [RFC7242] |
4 | TCPCLv4 | This specification. |
5-255 | Unassigned |
EDITOR NOTE: sub-registry to-be-created upon publication of this specification.¶
IANA will create, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version 4 Session Extension Types" and initialize it with the contents of Table 12. The registration procedure is Expert Review within the lower range 0x0001--0x7FFF. Values in the range 0x8000--0xFFFF are reserved for use on private networks for functions not published to the IANA.¶
Specifications of new session extension types need to define the encoding of the Item Value data as well as any meaning or restriction on the number of or order of instances of the type within an extension item list. Specifications need to define how the extension functions when no instance of the new extension type is received during session negotiation.¶
Expert(s) are encouraged to be biased towards approving registrations unless they are abusive, frivolous, or actively harmful (not merely aesthetically displeasing, or architecturally dubious).¶
Code | Session Extension Type |
---|---|
0x0000 | Reserved |
0x0001--0x7FFF | Unassigned |
0x8000--0xFFFF | Private/Experimental Use |
EDITOR NOTE: sub-registry to-be-created upon publication of this specification.¶
IANA will create, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version 4 Transfer Extension Types" and initialize it with the contents of Table 13. The registration procedure is Expert Review within the lower range 0x0001--0x7FFF. Values in the range 0x8000--0xFFFF are reserved for use on private networks for functions not published to the IANA.¶
Specifications of new transfer extension types need to define the encoding of the Item Value data as well as any meaning or restriction on the number of or order of instances of the type within an extension item list. Specifications need to define how the extension functions when no instance of the new extension type is received in a transfer.¶
Expert(s) are encouraged to be biased towards approving registrations unless they are abusive, frivolous, or actively harmful (not merely aesthetically displeasing, or architecturally dubious).¶
Code | Transfer Extension Type |
---|---|
0x0000 | Reserved |
0x0001 | Transfer Length Extension |
0x0002--0x7FFF | Unassigned |
0x8000--0xFFFF | Private/Experimental Use |
EDITOR NOTE: sub-registry to-be-created upon publication of this specification.¶
IANA will create, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version 4 Message Types" and initialize it with the contents of Table 14. The registration procedure is RFC Required within the lower range 0x01--0xEF. Values in the range 0xF0--0xFF are reserved for use on private networks for functions not published to the IANA.¶
Specifications of new message types need to define the encoding of the message data as well as the purpose and relationship of the new message to existing session/transfer state within the baseline message sequencing. The use of new message types need to be negotiated between TCPCL entities within a session (using the session extension mechanism) so that the receiving entity can properly decode all message types used in the session.¶
Expert(s) are encouraged to favor new session/transfer extension types over new message types. TCPCL messages are not self-delimiting, so care must be taken in introducing new message types. If an entity receives an unknown message type the only thing that can be done is to send a MSG_REJECT and close the TCP connection; not even a clean termination can be done at that point.¶
Code | Message Type |
---|---|
0x00 | Reserved |
0x01 | XFER_SEGMENT |
0x02 | XFER_ACK |
0x03 | XFER_REFUSE |
0x04 | KEEPALIVE |
0x05 | SESS_TERM |
0x06 | MSG_REJECT |
0x07 | SESS_INIT |
0x08--0xEF | Unassigned |
0xF0--0xFF | Private/Experimental Use |
EDITOR NOTE: sub-registry to-be-created upon publication of this specification.¶
IANA will create, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version 4 XFER_REFUSE Reason Codes" and initialize it with the contents of Table 15. The registration procedure is Specification Required within the lower range 0x00--0xEF. Values in the range 0xF0--0xFF are reserved for use on private networks for functions not published to the IANA.¶
Specifications of new XFER_REFUSE reason codes need to define the meaning of the reason and disambiguate it with pre-existing reasons. Each refusal reason needs to be usable by the receiving BP Agent to make retransmission or re-routing decisions.¶
Expert(s) are encouraged to be biased towards approving registrations unless they are abusive, frivolous, or actively harmful (not merely aesthetically displeasing, or architecturally dubious).¶
Code | Refusal Reason |
---|---|
0x00 | Unknown |
0x01 | Completed |
0x02 | No Resources |
0x03 | Retransmit |
0x04 | Not Acceptable |
0x05 | Extension Failure |
0x06 | Session Terminating |
0x07--0xEF | Unassigned |
0xF0--0xFF | Private/Experimental Use |
EDITOR NOTE: sub-registry to-be-created upon publication of this specification.¶
IANA will create, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version 4 SESS_TERM Reason Codes" and initialize it with the contents of Table 16. The registration procedure is Specification Required within the lower range 0x00--0xEF. Values in the range 0xF0--0xFF are reserved for use on private networks for functions not published to the IANA.¶
Specifications of new SESS_TERM reason codes need to define the meaning of the reason and disambiguate it with pre-existing reasons. Each termination reason needs to be usable by the receiving BP Agent to make re-connection decisions.¶
Expert(s) are encouraged to be biased towards approving registrations unless they are abusive, frivolous, or actively harmful (not merely aesthetically displeasing, or architecturally dubious).¶
Code | Termination Reason |
---|---|
0x00 | Unknown |
0x01 | Idle timeout |
0x02 | Version mismatch |
0x03 | Busy |
0x04 | Contact Failure |
0x05 | Resource Exhaustion |
0x06--0xEF | Unassigned |
0xF0--0xFF | Private/Experimental Use |
EDITOR NOTE: sub-registry to-be-created upon publication of this specification.¶
IANA will create, under the "Bundle Protocol" registry [IANA-BUNDLE], a sub-registry titled "Bundle Protocol TCP Convergence-Layer Version 4 MSG_REJECT Reason Codes" and initialize it with the contents of Table 17. The registration procedure is Specification Required within the lower range 0x01--0xEF. Values in the range 0xF0--0xFF are reserved for use on private networks for functions not published to the IANA.¶
Specifications of new MSG_REJECT reason codes need to define the meaning of the reason and disambiguate it with pre-existing reasons. Each rejection reason needs to be usable by the receiving TCPCL Entity to make message sequencing and/or session termination decisions.¶
Expert(s) are encouraged to be biased towards approving registrations unless they are abusive, frivolous, or actively harmful (not merely aesthetically displeasing, or architecturally dubious).¶
Code | Rejection Reason |
---|---|
0x00 | reserved |
0x01 | Message Type Unknown |
0x02 | Message Unsupported |
0x03 | Message Unexpected |
0x04--0xEF | Unassigned |
0xF0--0xFF | Private/Experimental Use |
This specification is based on comments on implementation of [RFC7242] provided from Scott Burleigh.¶
The areas in which changes from [RFC7242] have been made to existing headers and messages are:¶
The areas in which extensions from [RFC7242] have been made as new messages and codes are:¶