Delay-Tolerant Networking TCP Convergence Layer Protocol Version 4
RKF Engineering Solutions, LLC
7500 Old Georgetown RoadSuite 1275BethesdaMD20814-6198United States of AmericaBSipos@rkf-eng.com
University of California, Berkeley
Computer Science Division445 Soda HallBerkeleyCA94720-1776United States of Americademmer@cs.berkeley.edu
Aalto University
Department of Communications and NetworkingPO Box 13000Aalto02015Finlandott@in.tum.deQuebecQCCanadasimon@per.reau.lt
Transport
Delay Tolerant Networking
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 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"
.
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)
,
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
) 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.
This document describes the format of the protocol data units passed
between entities participating in TCPCL communications.
This document does not address:
The format of protocol data units of the Bundle Protocol, as those
are defined elsewhere in
.
This includes the concept of bundle fragmentation or bundle encapsulation.
The TCPCL transfers bundles as opaque data blocks.
Mechanisms for locating or identifying other bundle entities
(peers) within a network or across an internet.
The mapping of Node ID to potential convergence layer (CL)
protocol and network address
is left to implementation and configuration of the BP Agent
and its various potential routing strategies.
Logic for routing bundles along a path toward a bundle's endpoint.
This CL protocol is involved only in transporting bundles
between adjacent nodes in a routing sequence.
Policies or mechanisms for creating X.509 certificates;
provisioning, deploying, or accessing certificates and private keys;
deploying or accessing certificate revocation lists (CRLs);
or configuring security parameters on an individual entity or
across a network.
Uses of TLS which are not based on X.509 certificate
authentication (see )
or in which authentication of both entities is not possible
(see ).
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
when, and only when, they appear in all capitals, as shown here.
This section contains definitions specific to the TCPCL protocol.
Most significant byte first, a.k.a., big endian.
All of the integer encodings in this protocol SHALL be transmitted in
network byte order.
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:
A TCPCL Entity MAY actively initiate any number of TCPCL Sessions and should do so
whenever the entity is the initial transmitter of information to another entity in the network.
A TCPCL Entity MAY support zero or more passive listening elements that listen for
connection requests from other TCPCL Entities operating on other entities in the network.
A TCPCL Entity MAY passively initiate any number of TCPCL Sessions from requests received
by its passive listening element(s) if the entity uses such elements.
These relationships are illustrated in .
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 term Connection in this specification exclusively refers to
a TCP connection and any and all behaviors, sessions, and other
states associated with that TCP connection.
A TCPCL session (as opposed to a TCP connection) is a TCPCL
communication relationship between two TCPCL entities.
A TCPCL session operates within a single underlying TCP connection
and the lifetime of a TCPCL session is bound to the lifetime of
that TCP connection.
A TCPCL session is terminated when the TCP
connection ends, due either to one or both entities actively
closing the TCP connection or due to network errors causing
a failure of the TCP connection.
Within a single TCPCL session there are two possible transfer streams;
one in each direction, with one stream from each entity being the outbound
stream and the other being the inbound stream
(see ).
From the perspective of a TCPCL session, the two transfer streams
do not logically interact with each other.
The streams do operate over the same TCP connection and between the
same BP agents, so there are logical relationships at those layers
(message and bundle interleaving respectively).
For the remainder of this document, the term "session"
without the prefix "TCPCL" refers to a TCPCL session.
These are a set of
values used to affect the operation of the TCPCL for a given
session. The manner in which these parameters are conveyed
to the bundle entity and thereby to the TCPCL is implementation
dependent. However, the mechanism by which two entities
exchange and negotiate the values to be used for a given session
is described in
.
A Transfer stream is a uni-directional user-data path within a TCPCL Session.
Transfers sent over a transfer stream are serialized, meaning that one
transfer must complete its transmission prior to another transfer
being started over the same transfer stream.
At the stream layer there is no logical relationship between transfers
in that stream; it's only within the BP agent that transfers are
fully decoded as bundles.
Each uni-directional stream has a single sender entity and a single receiver entity.
This refers to the procedures and mechanisms
for conveyance of an individual bundle from one node to
another.
Each transfer within TCPCL is identified by a Transfer ID number
which is guaranteed to be unique only to a single direction
within a single Session.
A subset of a transfer of user data being communicated over a transfer stream.
A TCPCL session is idle while there is no transmission in-progress
in either direction.
While idle, the only messages being transmitted or received
are KEEPALIVE messages.
A TCPCL session is live while there is a transmission in-progress
in either direction.
The TCPCL uses numeric codes to encode specific reasons for individual
failure/error message types.
The relationship between connections, sessions, and streams is shown in .
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 node
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 allows a BP agent to preemptively attempt to establish
a TCPCL session with a peer entity.
Each session attempt can send a different set of session negotiation
parameters as directed by the BP agent.
The TCPCL allows a BP agent to preemptively terminate an established
TCPCL session with a peer entity.
The terminate request is on a per-session basis.
The TCPCL entity indicates to the BP agent when the session state changes.
The top-level session states indicated are:
A TCP connection is being established. This state only applies to the active entity.A TCP connection has been made (as either active or passive entity) and contact negotiation has begun.Contact negotiation has been completed (including possible TLS use) and session negotiation has begun.The session has been fully established and is ready for its first transfer.The entity sent SESS_TERM message and is in the ending state.The session has finished normal termination sequencing.The session ended without normal termination sequencing.
The TCPCL entity indicates to the BP agent when the live/idle sub-state of
the session changes.
This occurs only when the top-level session state is
"Established".
The session transitions from Idle to Live at the at the start
of a transfer in either transfer stream;
the session transitions from Live to Idle at the end of a
transfer when the other transfer stream does not have an ongoing
transfer.
Because TCPCL transmits serially over a TCP connection it suffers
from "head of queue blocking," so a transfer in either direction
can block an immediate start of a new transfer in the session.
The principal purpose of the TCPCL is to allow a BP agent to
transmit bundle data over an established TCPCL session.
Transmission request is on a per-session basis and the CL does not
necessarily perform any per-session or inter-session queueing.
Any queueing of transmissions is the obligation of the BP agent.
The TCPCL entity indicates to the BP agent when a bundle has been fully
transferred to a peer entity.
The TCPCL entity indicates to the BP agent on intermediate progress
of transfer to a peer entity.
This intermediate progress is at the granularity of each
transferred segment.
The TCPCL entity indicates to the BP agent on certain reasons for
bundle transmission failure, notably when the peer entity rejects
the bundle or when a TCPCL session ends before transfer success.
The TCPCL itself does not have a notion of transfer timeout.
The TCPCL entity indicates to the receiving BP agent just before any
transmission data is sent.
This corresponds to reception of the XFER_SEGMENT message with
the START flag of 1.
The TCPCL entity allows a BP agent to interrupt an individual transfer
before it has fully completed (successfully or not).
Interruption can occur any time after the reception is initialized.
The TCPCL entity indicates to the BP agent when a bundle has been fully
transferred from a peer entity.
The TCPCL entity indicates to the BP agent on intermediate progress
of transfer from the peer entity.
This intermediate progress is at the granularity of each
transferred segment.
Intermediate reception indication allows a BP agent the chance
to inspect bundle header contents before the entire bundle is
available, and thus supports the
"Reception Interruption" capability.
The TCPCL entity indicates to the BP agent on certain reasons for
reception failure, notably when the local entity rejects an attempted
transfer for some local policy reason or when a TCPCL session
ends before transfer success.
The TCPCL itself does not have a notion of transfer timeout.
First, one node establishes a TCPCL session to the other by
initiating a TCP connection in accordance with
.
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 ).
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 node 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 node to send
acknowledgment (XFER_ACK) messages as bundle data segments arrive. The
rationale behind these acknowledgments is to enable the sender node
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 node live-ness information during otherwise
message-less time intervals.
A SESS_TERM message is used to initiate the ending of a TCPCL session
(see ).
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
).
Regardless of the reason, session termination is initiated by one
of the entities and responded-to by the other as illustrated by
and
.
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 .
Notes on Established Session states:
Session "Live" means transmitting or receiving over a transfer stream.Session "Idle" means no transmission/reception over a transfer stream.Session "Ending" means no new transfers will be allowed.
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
and respectively which both share the data
validation and negotiation of the
Processing of Contact Header "[PCH]" activity of
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.
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
and respectively which both share the data
validation and negotiation of .
The validation here includes certificate validation and authentication
when TLS is used for the session.
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
and
respectively.
Notes on transfer sending:
Pipelining of transfers can occur when the sending entity begins
a new transfer while in the "Waiting for Ack" state.
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.
Each TCPCL session allows a negotiated transfer segmentation polcy
to be applied in each transfer direction.
A receiving node can set the Segment MRU in its SESS_INIT message to
determine the largest acceptable segment size, and a transmitting
node 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.
For a simple network expected to exchange relatively small bundles,
the Segment MRU can be set to be identical to the Transfer MRU which
indicates that all transfers can be sent with a single data
segment (i.e., no actual segmentation).
If the network is closed and all transmitters are known to follow
a single-segment transfer policy, then receivers can avoid the
necessity of segment reassembly.
Because this CL operates over a TCP stream, which suffers from
a form of head-of-queue blocking between messages, while one
node is transmitting a single XFER_SEGMENT message it is not
able to transmit any XFER_ACK or XFER_REFUSE for any associated
received transfers.
In situations where the maximum message size is desired to be
well-controlled, the
Segment MRU can be set to the largest acceptable size (the
message size less XFER_SEGMENT header size) and transmitters
can always segment a transfer into maximum-size chunks no larger
than the Segment MRU.
This guarantees that any single XFER_SEGMENT will not monopolize
the TCP stream for too long, which would prevent outgoing
XFER_ACK and XFER_REFUSE associated with received transfers.
Even after negotiation of a Segment MRU for each receiving node,
the actual transfer segmentation only needs to guarantee than
any individual segment is no larger than that MRU.
In a situation where TCP throughput is dynamic, the transfer
segmentation size can also be dynamic in order to control
message transmission duration.
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 node.
Additionally, future header 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 node 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.
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 10 minutes (600 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
.
Because the TCPCL protocol version in use is part of the initial
Contact Header, nodes using TCPCL version 4 can coexist on a network
with nodes using earlier TCPCL versions (with some negotiation needed
for interoperation as described in
).
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 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 .
Upon receipt of the Contact Header, both entities perform the validation
and negotiation procedures defined in
.
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:
See
for details on the use of each
of these Contact Header fields.
The fields of the Contact Header are:
A four-octet field that always contains the octet sequence 0x64
0x74 0x6E 0x21, i.e., the text string "dtn!" in US-ASCII
(and UTF-8).
A one-octet field value containing the value 4 (current
version of the TCPCL).
A one-octet field of single-bit flags, interpreted according to the
descriptions in
.
All reserved header flag bits SHALL be set to 0 by the sender.
All reserved header flag bits SHALL be ignored by the receiver.
NameCodeDescriptionCAN_TLS0x01If bit is set, indicates that the sending peer is capable of TLS security.Reservedothers
Upon reception of the Contact Header, each node 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 node 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 node 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.
Both Contact Headers of a successful contact negotiation
have identical TCPCL Version numbers as described above.
Only upon response of a Contact Header from the passive entity is
the TCPCL protocol version established and session negotiation begun.
Negotiation of the Enable TLS parameter is performed by taking
the logical AND of the two Contact Headers' CAN_TLS flags.
A local security policy is then applied to determine of the negotiated value
of Enable TLS is acceptable.
It can be a reasonable security policy to require or disallow
the use of TLS depending upon the desired network flows.
Because this state is negotiated over an unsecured medium,
there is a risk of a TLS Stripping as described in .
If the Enable TLS state is unacceptable, the entity SHALL terminate
the session with a reason code of "Contact Failure".
Note that this contact failure reason is different than a failure
of TLS handshake or TLS authentication
after an agreed-upon and acceptable Enable TLS state.
If the negotiated Enable TLS value is true and acceptable then
TLS negotiation feature (described in
) begins
immediately following the Contact Header exchange.
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
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 the 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.
The public key infrastructure (PKI)
Certificate Authority (CA) or authorities available
within a network are likely not controlled by the certificate
end users and CA policies vary widely between networks and
PKI management tools.
For this reason, the TCPCL defines a prioritized list of what
a certificate can identify about a TCPCL entity:
The ideal certificate identity is the Node ID of the entity
using the NODE-ID definition below.
When the Node ID is identified, there is no need for any
lower-level identification to take place.
If CA policy forbids a certificate to contain an arbitrary
NODE-ID but allows a DNS-ID to be identified then one or more
stable host names can be identified in the certificate.
The use of wildcard DNS-ID is discouraged due to the
complex rules for matching and dependence on implementation
support for wildcard matching.
If no stable host name is available but a stable network address
is available and CA policy allows a certificate to contain a
NETWORK-ID (as defined below)
then one or more network addresses can be identified in the
certificate.
There is no wildcard-type address matching defined, so this
is the least robust
When only a DNS-ID or NETWORK-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 peer
which authenticates its Node ID regardless of an authenticated
host name or address, and only consider the host/address
authentication in the absence of an authenticated 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 and the URI schemes of
the IANA "Bundle Protocol URI Scheme Type" registry.
This is similar to the URI-ID of 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
and scheme-based normalization of
those schemes specified in .
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 NETWORK-ID of a certificate as
being the subjectAltName entry of type iPAddress whose value is encoded
according to .
The use of TLS is negotiated using the Contact Header as described
in .
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 .
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
.
The parameters within each TLS negotiation are implementation dependent but
any TCPCL node SHALL follow all recommended practices of
BCP 195 , 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):
When a resolved host name was used to establish the TCP connection,
the TLS ClientHello SHOULD include a Server Name Indication (SNI)
in accordance with
containing that host name (of the passive entity) which was resolved.
Note: The SNI text is the network-layer name for the passive entity,
which is not the Node ID of that entity.
The passive entity SHALL supply a certificate within the
TLS handshake to allow authentication of its side of the session.
Unless prohibited by CA policy,
the passive entity certificate SHALL contain a NODE-ID
which authenticates the Node ID of the peer.
When assigned a stable host name,
the passive entity certificate SHOULD contain a DNS-ID
which authenticates that (fully qualified) name.
When assigned a stable network address,
the passive entity certificate MAY contain a NETWORK-ID
which authenticates that address.
The passive entity MAY use the SNI host name to choose an appropriate
server-side certificate which authenticates that host name.
During TLS handshake, the passive entity SHALL request a
client-side certificate.
The active entity SHALL supply a certificate chain within the
TLS handshake to allow authentication of its side of the session.
Unless prohibited by CA policy,
the active entity certificate SHALL contain a NODE-ID
which authenticates the Node ID of the peer.
When assigned a stable host name,
the active entity certificate SHOULD contain a DNS-ID
which authenticates that (fully qualified) name.
When assigned a stable network address,
the active entity certificate MAY contain a NETWORK-ID
which authenticates that address.
All certificates supplied during TLS handshake SHALL conform to
,
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 X.509 certificates exchanged during the TLS handshake, each
of the entities can attempt to authenticate its peer Node ID directly
or authenticate the peer host name or network address.
The Node ID exchanged in the Session Initialization is likely to
be used by the BP agent for making transfer and routing decisions,
so attempting Node ID validation is required
while attempting host name validation is optional.
The logic for attempting validation is separate from the
logic for handling the result of validation, which is based on
local security policy.
By using the SNI host name (see ) 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 host 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 host names and/or Node IDs being hosted or by several
certificates each authenticating a single host name and/or Node ID,
using the SNI value from the peer to select which certificate to use.
Any certificate received during TLS handshake SHALL be validated
up to one or more trusted CA certificates.
If certificate validation fails or if security policy disallows
a certificate for any reason, the entity SHALL terminate the session
(with a reason code of "Contact Failure").
Either during or immediately after the TLS handshake,
the active entity SHALL attempt to authenticate the host name
(of the passive entity) used to initiate the TCP connection
using any DNS-ID of the peer certificate.
If host name validation fails (including failure
because the certificate does not contain any DNS-ID) and
security policy disallows an unauthenticated host,
the entity SHALL terminate the session
(with a reason code of "Contact Failure").
Either during or immediately after the TLS handshake,
the active entity SHALL attempt to authenticate the IP address
of the other side of the TCP connection using any NETWORK-ID of
the peer certificate.
Either during or immediately after the TLS handshake,
the passive entity SHALL attempt to authenticate the IP address of
the other side of the TCP connection using any NETWORK-ID of
the peer certificate.
If host address validation fails (including failure
because the certificate does not contain any NETWORK-ID) and
security policy disallows an unauthenticated host,
the entity SHALL terminate the session
(with a reason code of "Contact Failure").
Immediately before Session Parameter Negotiation, each side of
the session SHALL perform Node ID validation of its peer as
described below.
Node ID validation SHALL succeed if the associated certificate
includes a NODE-ID whose value matches the Node ID of the TCPCL entity.
If Node ID validation fails (including failure because
the certificate does not contain any NODE-ID) and
security policy disallows an unauthenticated 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
below.
In this example the active peer terminates the session but
termination can be initiated from either peer.
After the initial exchange of a Contact Header, all messages
transmitted over the session are identified by a one-octet header
with the following structure:
The message header fields are as follows:
Indicates the type of the message as per
below.
Encoded values are listed in
.
NameCodeDescriptionSESS_INIT0x07
Contains the session parameter inputs from one of the entities,
as described in .
SESS_TERM0x05
Indicates that one of the entities participating in the session wishes to cleanly terminate the session, as described in
.
XFER_SEGMENT0x01
Indicates the transmission of a segment of bundle data, as described in
.
XFER_ACK0x02
Acknowledges reception of a data segment, as described in
.
XFER_REFUSE0x03
Indicates that the transmission of the current bundle SHALL be stopped, as described in
.
KEEPALIVE0x04
Used to keep TCPCL session active, as described in
.
MSG_REJECT0x06
Contains a TCPCL message rejection, as described in
.
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 .
The format of a SESS_INIT message is as follows in
.
The fields of the SESS_INIT message are:
A 16-bit unsigned integer indicating the minimum interval, in seconds,
to negotiate the Session Keepalive using the
method of .
A 64-bit unsigned integer indicating the largest allowable single-segment
data payload size to be received in this session.
Any XFER_SEGMENT sent to this peer SHALL have a data payload no longer
than the peer's Segment MRU.
The two entities of a single session MAY have different Segment MRUs, and
no relation between the two is required.
A 64-bit unsigned integer indicating the largest allowable total-bundle
data size to be received in this session.
Any bundle transfer sent to this peer SHALL have a Total Bundle Length
payload no longer than the peer's Transfer MRU.
This value can be used to perform proactive bundle fragmentation.
The two entities of a single session MAY have different Transfer MRUs, and
no relation between the two is required.
Together these fields represent a variable-length text string.
The Node ID Length is a 16-bit unsigned integer indicating the number of
octets of Node ID Data to follow.
A zero-length Node ID SHALL be used to indicate the lack of
Node ID rather than a truly empty Node ID.
This case allows an entity to avoid exposing Node ID information on an
untrusted network.
A non-zero-length Node ID Data SHALL contain the UTF-8 encoded
Node ID of the Entity which sent the SESS_INIT message.
Every Node ID SHALL be a URI consistent with the requirements
of and the URI schemes of
the IANA "Bundle Protocol URI Scheme Type" registry.
The Node ID itself can be authenticated as described in
.
Together these fields represent protocol extension data
not defined by this specification.
The Session Extension Length is the total number of octets to follow which
are used to encode the Session Extension Item list.
The encoding of each Session Extension Item is within a consistent data
container as described in
.
The full set of Session Extension Items apply for the duration of
the TCPCL session to follow.
The order and multiplicity of these Session Extension Items is
significant, as defined in the associated type specification(s).
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 node
(expressed in the SESS_INIT that it received from the peer).
The negotiated parameters defined by this specification are
described in the following paragraphs.
The maximum transmit unit (MTU) for whole transfers and individual segments
are identical to the Transfer MRU and Segment MRU, respectively,
of the received Contact Header.
A transmitting peer can send individual segments with any size smaller
than the Segment MTU, depending on local policy, dynamic network
conditions, etc.
Determining the size of each transmitted segment is an implementation
matter.
If either the Transfer MRU or Segment MRU is unacceptable, the entity
SHALL terminate the session with a reason code of
"Contact Failure".
Negotiation of the Session Keepalive parameter is performed by taking
the minimum of the two Contact Headers' Keepalive Interval values.
The Session Keepalive interval is a parameter for the behavior described
in .
If the Session Keepalive interval is unacceptable, the entity
SHALL terminate the session with a reason code of
"Contact Failure".
Note: a negotiated Session Keepalive of zero indicates that
KEEPALIVEs are disabled.
Once this process of parameter negotiation is completed (which includes a
possible completed TLS handshake of the connection to use TLS),
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
.
The fields of the Session Extension Item are:
A one-octet field containing generic bit flags about the Item,
which are listed in
.
All reserved header flag bits SHALL be set to 0 by the sender.
All reserved header flag bits SHALL be ignored by the receiver.
If a TCPCL entity receives a Session Extension Item with an unknown Item Type
and the CRITICAL flag of 1, the entity SHALL close the
TCPCL session with SESS_TERM reason code of "Contact Failure".
If the CRITICAL flag is 0, an entity SHALL skip over and ignore
any item with an unknown Item Type.
A 16-bit unsigned integer field containing the type of the extension item.
This specification does not define any extension types directly, but does
create an IANA registry for such codes
(see ).
A 16-bit unsigned integer field containing the number of Item Value octets
to follow.
A variable-length data field which is interpreted according to the
associated Item Type.
This specification places no restrictions on an extension's use of
available Item Value data.
Extension specifications SHOULD avoid the use of large data lengths,
as no bundle transfers can begin until
the full extension data is sent.
NameCodeDescriptionCRITICAL0x01If bit is set, indicates that the receiving peer must handle the extension item.Reservedothers
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
,
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
) 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
) 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
.
The fields of the MSG_REJECT message are:
A one-octet refusal reason code interpreted according to the
descriptions in
.
The Rejected Message Header is a copy of the Message Header to which the
MSG_REJECT message is sent as a response.
NameCodeDescriptionMessage Type Unknown0x01A message was received with
a Message Type code unknown to the TCPCL node.Message Unsupported0x02A message was received but
the TCPCL entity cannot comply with the message contents.Message Unexpected0x03A 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 node 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 ).
The choice of the length to use for segments is an implementation matter,
but each segment MUST NOT be larger than the receiving node's maximum
receive unit (MRU)
(see the field Segment MRU of ).
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 node SHALL be unique within a single TCPCL session.
Upon exhaustion of the entire 64-bit Transfer ID space, the sending node
SHALL terminate the session with SESS_TERM reason code "Resource Exhaustion".
For bidirectional bundle transfers, a TCPCL node 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 ),
in the absence of any other mechanism for
generating Transfer IDs an entity SHALL use the following algorithm:
The initial Transfer ID from each node 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
.
The fields of the XFER_SEGMENT message are:
A one-octet field of single-bit flags, interpreted according to the
descriptions in
.
All reserved header flag bits SHALL be set to 0 by the sender.
All reserved header flag bits SHALL be ignored by the receiver.
A 64-bit unsigned integer identifying the transfer being made.
Together these fields represent protocol extension data
for this specification.
The Transfer Extension Length and Transfer Extension Item fields
SHALL only be present when the 'START' flag is set to 1 on the message.
The Transfer Extension Length is the total number of octets to follow which
are used to encode the Transfer Extension Item list.
The encoding of each Transfer Extension Item is within a consistent data
container as described in
.
The full set of transfer extension items apply only to the
associated single transfer.
The order and multiplicity of these transfer extension items is
significant, as defined in the associated type specification(s).
A 64-bit unsigned integer indicating the number of octets in the
Data contents to follow.
The variable-length data payload of the message.
NameCodeDescriptionEND0x01If bit is set, indicates that this is the last segment of the transfer.START0x02If bit is set, indicates that this is the first segment of the transfer.Reservedothers
The flags portion of the message contains two flag
values in the two low-order bits, denoted 'START' and 'END' in
.
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 node 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 node transmits acknowledgments of reception of data
segments.
The format of an XFER_ACK message follows in
.
The fields of the XFER_ACK message are:
A one-octet field of single-bit flags, interpreted according to the
descriptions in
.
All reserved header flag bits SHALL be set to 0 by the sender.
All reserved header flag bits SHALL be ignored by the receiver.
A 64-bit unsigned integer identifying the transfer being acknowledged.
A 64-bit unsigned integer indicating the total number of octets in the
transfer which are being acknowledged.
A receiving TCPCL node 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 node 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 node 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 node 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
node (which is supposed to represent a firm limitation of what the
node 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
.
The fields of the XFER_REFUSE message are:
A one-octet refusal reason code interpreted according to the
descriptions in
.
A 64-bit unsigned integer identifying the transfer being refused.
NameCodeDescriptionUnknown0x00Reason for refusal is unknown or not specified.Completed0x01The receiver already has the complete bundle. The sender MAY consider the bundle as completely received.No Resources0x02The receiver's resources are exhausted. The sender SHOULD apply reactive bundle fragmentation before retrying.Retransmit0x03The receiver has encountered a problem that requires the bundle to be retransmitted in its entirety.Not Acceptable0x04Some 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 Failure0x05A failure processing the Transfer Extension Items has occurred.
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
.
The fields of the Transfer Extension Item are:
A one-octet field containing generic bit flags about the Item,
which are listed in
.
All reserved header flag bits SHALL be set to 0 by the sender.
All reserved header flag bits SHALL be ignored by the receiver.
If a TCPCL node receives a Transfer Extension Item with an unknown Item Type
and the CRITICAL flag is 1, the entity SHALL refuse the transfer
with an XFER_REFUSE reason code of "Extension Failure".
If the CRITICAL flag is 0, an entity SHALL skip over and ignore
any item with an unknown Item Type.
A 16-bit unsigned integer field containing the type of the extension item.
This specification creates an IANA registry for such codes
(see ).
A 16-bit unsigned integer field containing the number of Item Value octets
to follow.
A variable-length data field which is interpreted according to the
associated Item Type.
This specification places no restrictions on an extension's use of
available Item Value data.
Extension specifications SHOULD avoid the use of large data lengths,
as the associated transfer cannot begin until
the full extension data is sent.
NameCodeDescriptionCRITICAL0x01If bit is set, indicates that the receiving peer must handle the extension item.Reservedothers
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 node 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
.
The fields of the Transfer Length extension are:
A 64-bit unsigned integer indicating the size of the data-to-be-transferred.
The Total Length field SHALL be treated as authoritative by the receiver.
If, for whatever reason, the actual total length of bundle data
received differs from the value indicated by the
Total Length value, the receiver SHALL treat the transmitted data as invalid
and send an XFER_REFUSE with a Reason Code of "Not Acceptable".
This section describes the procedures for terminating a TCPCL session.
The purpose of terminating a session is to allow transfers to complete
before the session 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 node 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.
Instead of following a clean termination sequence, after transmitting a
SESS_TERM message an entity MAY immediately close
the associated TCP connection.
When performing an unclean termination, a receiving node 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.
When performing an unclean termination, a transmitting node SHALL treat
either sending or receiving a SESS_TERM message
(i.e., before the final acknowledgment) as a failure of the transfer.
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.
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
.
The fields of the SESS_TERM message are:
A one-octet field of single-bit flags, interpreted according to the
descriptions in
.
All reserved header flag bits SHALL be set to 0 by the sender.
All reserved header flag bits SHALL be ignored by the receiver.
A one-octet refusal reason code interpreted according to the
descriptions in
.
NameCodeDescriptionREPLY0x01If bit is set, indicates that this message is an acknowledgement of an earlier SESS_TERM message.ReservedothersNameCodeDescriptionUnknown0x00A termination reason is not available.Idle timeout0x01The session is being closed due to idleness.Version mismatch0x02The node cannot conform to the specified TCPCL protocol version.Busy0x03The node is too busy to handle the current session.Contact Failure0x04The node cannot interpret or negotiate a Contact Header or SESS_INIT option.Resource Exhaustion0x05The node has run into some resource limit and cannot continue the session.
A session termination MAY occur 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.
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 the content of the Session Extension Items data disagrees with the
Session Extension Length (i.e., the last Item claims to use more octets
than are present in the Session Extension Length), the reception of the
SESS_INIT is considered to have failed.
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 ending
idle sessions, if they are to be closed at all, is an
implementation and configuration matter.
If there is a configured time to close idle links and if no TCPCL messages
(other than KEEPALIVE messages) has been received for at least
that amount of time, then either node MAY terminate the session by
transmitting a SESS_TERM message indicating the reason code of "Idle
timeout" (as described in
).
[NOTE to the RFC Editor: please remove this section before
publication, as well as the reference to
and
.]
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
.
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
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 .
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 ).
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
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 ).
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 node 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 .
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.
The negotiated use of TLS is identical behavior to STARTTLS use in
and .
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 .
It is up to security policies within each TCPCL node to ensure that the
negotiated TLS ciphersuite meets transport security requirements.
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 , 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) 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
.
When permitted by the negotiated TLS version
(see ), it is advisable to take
advantage of session key updates to avoid those limits.
When key updates are not possible, renegotiation of the TLS connection or
establishing new TCPCL/TLS session
are alternatives to limit session key use.
The certificates exchanged by TLS enable authentication of peer
host 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 host name or Node ID of the peer.
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
bind the host name or node ID based on the contents of the certificate.
The host name validation is a weaker form of authentication, because
even if a peer is operating on an authenticated network host 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.
Trusting an authenticated host name can be feasible on a network
secured by a private CA but is not advisable on the Internet when
using a variety of public CAs.
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 host name validation if certificates
can be guaranteed to validate the peer's Node ID.
In circumstances where certificates can only be issued to network
host 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 host 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 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 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 X.509 PKI
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 authenticate both peer entities of each
TCPCL session is outside of the scope of this document but does have similar
properties to the opportunistic security model of .
In environments where PKI is unavailable,
alternate uses of TLS which do not require certificates such as
pre-shared key (PSK) authentication and the use of
raw public keys
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
.
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 ,
TCP port number 4556 has been previously assigned as the default port for the
TCP convergence layer in .
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 ), so there is no
ambiguity about what protocol is being used.
The related assignments for UDP and DCCP port 4556 (both registered
by ) are unchanged.
ParameterValueService Name:dtn-bundleTransport Protocol(s):TCPAssignee:IESG <iesg@ietf.org>Contact:IESG <iesg@ietf.org>Description:DTN Bundle TCP CL ProtocolReference:This specification.Port Number:4556
IANA has created, under the "Bundle Protocol" registry
, 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.
ValueDescriptionReference0Reserved1Reserved2Reserved3TCPCL4TCPCLv4This specification.5-255UnassignedEDITOR NOTE: sub-registry to-be-created upon publication of this specification.
IANA will create, under the "Bundle Protocol" registry
, a sub-registry titled
"Bundle Protocol TCP Convergence-Layer Version 4 Session Extension Types"
and initialize it with the contents of
.
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).
CodeSession Extension Type0x0000Reserved0x0001--0x7FFFUnassigned0x8000--0xFFFFPrivate/Experimental UseEDITOR NOTE: sub-registry to-be-created upon publication of this specification.
IANA will create, under the "Bundle Protocol" registry
, a sub-registry titled
"Bundle Protocol TCP Convergence-Layer Version 4 Transfer Extension Types"
and initialize it with the contents of
.
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).
CodeTransfer Extension Type0x0000Reserved0x0001Transfer Length Extension0x0002--0x7FFFUnassigned0x8000--0xFFFFPrivate/Experimental UseEDITOR NOTE: sub-registry to-be-created upon publication of this specification.
IANA will create, under the "Bundle Protocol" registry
, a sub-registry titled
"Bundle Protocol TCP Convergence-Layer Version 4 Message Types"
and initialize it with the contents of
.
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.
CodeMessage Type0x00Reserved0x01XFER_SEGMENT0x02XFER_ACK0x03XFER_REFUSE0x04KEEPALIVE0x05SESS_TERM0x06MSG_REJECT0x07SESS_INIT0x08--0xEFUnassigned0xF0--0xFFPrivate/Experimental UseEDITOR NOTE: sub-registry to-be-created upon publication of this specification.
IANA will create, under the "Bundle Protocol" registry
, a sub-registry titled
"Bundle Protocol TCP Convergence-Layer Version 4 XFER_REFUSE Reason Codes"
and initialize it with the contents of
.
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).
CodeRefusal Reason0x00Unknown0x01Completed0x02No Resources0x03Retransmit0x04Not Acceptable0x05Extension Failure0x06--0xEFUnassigned0xF0--0xFFPrivate/Experimental UseEDITOR NOTE: sub-registry to-be-created upon publication of this specification.
IANA will create, under the "Bundle Protocol" registry
, a sub-registry titled
"Bundle Protocol TCP Convergence-Layer Version 4 SESS_TERM Reason Codes"
and initialize it with the contents of
.
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).
CodeTermination Reason0x00Unknown0x01Idle timeout0x02Version mismatch0x03Busy0x04Contact Failure0x05Resource Exhaustion0x06--0xEFUnassigned0xF0--0xFFPrivate/Experimental UseEDITOR NOTE: sub-registry to-be-created upon publication of this specification.
IANA will create, under the "Bundle Protocol" registry
, a sub-registry titled
"Bundle Protocol TCP Convergence-Layer Version 4 MSG_REJECT Reason Codes"
and initialize it with the contents of
.
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).
CodeRejection Reason0x00reserved0x01Message Type Unknown0x02Message Unsupported0x03Message Unexpected0x04--0xEFUnassigned0xF0--0xFFPrivate/Experimental Use
This specification is based on comments on implementation of
provided from Scott Burleigh.
Service Name and Transport Protocol Port Number RegistryIANABundle ProtocolIANALimits on Authenticated Encryption Use in TLSTCPCL Example Implementation
RKF Engineering Solutions, LLC
The areas in which changes from
have been made to existing headers and messages are:
Split Contact Header into pre-TLS protocol negotiation and SESS_INIT parameter negotiation. The Contact Header is now fixed-length.Changed Contact Header content to limit number of negotiated options.Added session option to negotiate maximum segment size (per each direction).Renamed "Endpoint ID" to "Node ID" to conform with BPv7 terminology.Added session extension capability.Added transfer extension capability. Moved transfer total length into an extension item.Defined new IANA registries for message / type / reason codes to allow renaming some codes for clarity.Segments of all new IANA registries are reserved for private/experimental use.Expanded Message Header to octet-aligned fields instead of bit-packing.Added a bundle transfer identification number to all bundle-related messages (XFER_SEGMENT, XFER_ACK, XFER_REFUSE).Use flags in XFER_ACK to mirror flags from XFER_SEGMENT.Removed all uses of SDNV fields and replaced with fixed-bit-length (network byte order) fields.Renamed SHUTDOWN to SESS_TERM to deconflict term "shutdown" related to TCP connections.Removed the notion of a re-connection delay parameter.
The areas in which extensions from
have been made as new messages and codes are:
Added contact negotiation failure SESS_TERM reason code.Added MSG_REJECT message to indicate an unknown or unhandled message was received.Added TLS connection security mechanism.Added Resource Exhaustion SESS_TERM reason code.