Internet-Draft | DTLS in SCTP | June 2023 |
Westerlund, et al. | Expires 30 December 2023 | [Page] |
This document defines a usage of Datagram Transport Layer Security (DTLS) 1.2 or 1.3 to protect the content of Stream Control Transmission Protocol (SCTP) packets using the framework provided by the SCTP CRYPTO chunk which we name DTLS in SCTP. DTLS in SCTP provides encryption, source authentication, integrity and replay protection for the SCTP association with mutual authentication of the peers. The specification is also targeting very long-lived sessions of weeks and months and supports mutual re-authentication and rekeying with ephemeral key exchange. This is intended as an alternative to using DTLS/SCTP (RFC 6083) and SCTP-AUTH (RFC 4895).¶
This note is to be removed before publishing as an RFC.¶
Status information for this document may be found at https://datatracker.ietf.org/doc/draft-westerlund-tsvwg-sctp-crypto-dtls/.¶
Discussion of this document takes place on the Transport Area Working Group (tsvwg) Working Group mailing list (mailto:tsvwg@ietf.org), which is archived at https://mailarchive.ietf.org/arch/browse/tsvwg/. Subscribe at https://www.ietf.org/mailman/listinfo/tsvwg/.¶
Source for this draft and an issue tracker can be found at https://github.com/gloinul/draft-westerlund-tsvwg-sctp-crypto-dtls.¶
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 30 December 2023.¶
Copyright (c) 2023 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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
This document describes the usage of the Datagram Transport Layer Security (DTLS) protocol, as defined in DTLS 1.2 [RFC6347], and DTLS 1.3 [RFC9147], as protection engine in the Stream Control Transmission Protocol (SCTP), as defined in [RFC9260] with SCTP CRYPTO chunk [I-D.westerlund-tsvwg-sctp-crypto-chunk]. This specification is intended as an alternative to DTLS/SCTP [RFC6083] and usage of SCTP-AUTH [RFC4895].¶
This specification provides mutual authentication of endpoints, data confidentiality, data origin authentication, data integrity protection, and data replay protection of SCTP packets. Ensuring these security services to the application and its upper layer protocol over SCTP. Thus, it allows client/server applications to communicate in a way that is designed with communications privacy and preventing eavesdropping and detect tampering or message forgery.¶
Applications using DTLS in SCTP can use all currently existing transport features provided by SCTP and its extensions, in some cases with some limitations, as specified in [I-D.westerlund-tsvwg-sctp-crypto-chunk]. DTLS in SCTP supports:¶
DTLS in SCTP is a protection engine specification for the SCTP CRYPTO chunk [I-D.westerlund-tsvwg-sctp-crypto-chunk] that utilizes DTLS 1.2 or 1.3 for the security functions like key exchange, authentication, encryption, integrity protection, and replay protection. The basic functionalities and how things are related are described below.¶
In a SCTP association initiation where DTLS in SCTP is chosen as the protection engine for the CRYPTO chunk the DTLS handshake is exchanged encapsulated in plain DATA chunks with Protection Engine PPID (see section 10.6 of [I-D.westerlund-tsvwg-sctp-crypto-chunk]) until an initial DTLS connection has been established. If the DTLS handshake fails, the SCTP association is aborted. When the DTLS connection has been established PVALID chunks are exchanged to verify that no downgrade attack between different protection engines has occurred. To prevent manipulation, the PVALID chunks are protected by encapsulating them in DTLS protected CRYPTO chunks.¶
Assuming that the PVALID validation is successful the SCTP association is established and the Upper Layer Protocol (ULP) can start sending data over the SCTP association. From this point all chunks will be protected by encapsulating them in DTLS protected CRYPTO chunks. The SCTP chunks to be included in an SCTP packet are the plain text application data input to DTLS. The encrypted DTLS application data record is then encapsulated in the CRYPTO chunk and the packet is transmitted, see Section 7.¶
In the receiving SCTP endpoint each incoming SCTP packet on any of its interfaces and ports are matched to the SCTP association based on ports and VTAG in the common header. In that association context for the CRYPTO chunk there will exist reference to one or more DTLS connections used to protect the data. The DTLS connection actually used to protect this packet is identified by two DCI bits in the CRYPTO chunk's flags. Using the identified DTLS session the content of the CRYPTO chunk is attempted to be processed, including replay protection, decryption, and integrity checking. And if decryption and integrity verification was successful the produced plain text of one or more SCTP chunks are provided for normal SCTP processing in the identified SCTP association along with associated meta data such as path received on, original packet size, and ECN bits.¶
When mutual re-authentication or rekeying with ephemeral key exchange is needed or desired by either endpoint a new DTLS connection handshake is performed between the SCTP endpoints. A different DTLS Connection Index (DCI) than currently used among the CRYPTO chunk flags are used to indicate that this is a new handshake. When the handshake has completed the DTLS in SCTP implementation can simply switch to use this DTLS connection to protect the plain text payload. After a short while (no longer than 2 min) to enable any outstanding packets to drain from the network path between the endpoints the old DTLS connection can be terminated.¶
The DTLS connection is free to send any alert, handshake message, or other non-application data to its peer at any point in time. Thus, enabling DTLS 1.3 Key Updates for example. All non-application data SHOULD be sent by means of SCTP DATA chunks with Protection Engine PPID as specified in [I-D.westerlund-tsvwg-sctp-crypto-chunk].¶
DTLS in SCTP has a number of properties that are attractive.¶
DTLS/SCTP as defined by [I-D.ietf-tsvwg-dtls-over-sctp-bis] has several important differences most to the benefit of DTLS in SCTP. This section reviews these differences.¶
There are several significant differences in regard to implementation between the two realizations.¶
DTLS in SCTP can use a DTLS implementation that does not rely on features from outside of the core protocol, where DTLS/SCTP required a number of features as listed below:¶
The conclusion of these implementation details is that where DTLS in SCTP can use existing DTLS implementations, including OpenSSL's DTLS 1.2 implementation. It is not known if any DTLS stack exist that fully support the requirements in DTLS/SCTP. It is expected that a DTLS/SCTP implementation will have to also extend some DTLS implementation.¶
This document uses the following terms:¶
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 identifies how the extension described in this document is identified in the Crypto Chunk and its negotiation [I-D.westerlund-tsvwg-sctp-crypto-chunk].¶
This document specifies the adoption of DTLS as protection engine for SCTP Crypto Chunks for DTLS1.2 and DTLS1.3¶
The following table applies.¶
VALUE | DTLS VERSION | REFERENCE |
---|---|---|
0 | DTLS 1.2 | RFC-To-Be |
1 | DTLS 1.3 | RFC-To-Be |
The values specified above shall be used in the Protected Association parameter as protection engines as specified in [I-D.westerlund-tsvwg-sctp-crypto-chunk] and are registered with IANA below in Section 11.1.¶
DTLS in SCTP uses the CRYPTO chunk in the following way. Fields not discussed are used as specified in [I-D.westerlund-tsvwg-sctp-crypto-chunk].¶
DTLS Connection Index is the lower two bits of an DTLS Connection Index counter. This is a counter implemented in DTLS in SCTP that is used to identify which DTLS connection instance that is capable of processing any received packet. This counter is recommended to be 64-bit to guarantee no lifetime issues for the SCTP Association.¶
Chunk Flag bits not currently used by DTLS in SCTP. They MUST be set to zero (0) and MUST be ignored on reception. They MAY be used in future updated specifications for DTLS in SCTP.¶
One or more DTLS records. In cases more than one DTLS record is included all DTLS records except the last MUST include a length field. Note that this matches what is specified in DTLS 1.3 [RFC9147] and DTLS 1.2 will always include the length field in each record.¶
There are a set of requirements stated in [I-D.westerlund-tsvwg-sctp-crypto-chunk] that need to be addressed in this specification, this section deals with those requirements and how they are met in the current specification.¶
The CRYPTO Chunk allows the protection engine to have inband or out-of-band key establishment. DTLS in SCTP uses inband key establishment, thus the DTLS handshake establishes shared keys with the remote peer. As soon as the SCTP State Machine enters PROTECTION PENDING state, DTLS is responsible for progressing to the PROTECTED state when DTLS handshake has completed. The DCI counter is initialized to the value zero that is used for the initial DTLS handshake.¶
When entering PROTECTION PENDING state, DTLS will start the handshake according to Section 6.1.¶
DTLS protection engine being initialized for a new SCTP association will set the DCI counter = 0, which implies a DCI field value of 0, for the initial DTLS connection. The DTLS handshake messages are transmitted from this endpoint to the peer using DATA chunks with the PPID value set to Protection Engine Protocol Identifier [I-D.westerlund-tsvwg-sctp-crypto-chunk].¶
When a successful handshake has been completed, DTLS protection engine will inform CRYPTO chunk Handler that will move SCTP State Machine into PROTECTED state.¶
In the PROTECTED state the currently active DTLS connection is used for protection operation of the payload of SCTP chunks in each packet per below specification. When necessary to meet requirements on periodic re-authentication of the peer and establishment of new forward secrecy keys a new parallel DTSL connection is established as further specified in Section 8.¶
When the SCTP association leaves the ESTABLISHED state per [RFC9260] to be shutdown the DTLS connection is kept and continues to protect the SCTP packet payloads through the shutdown process.¶
When the association reaches the CLOSED state as part of the SCTP association closing process all DTLS connections that existed are terminated without further transmissions, i.e. DTLS close_notify is not transmitted.¶
It's up to DTLS protection engine to manage the DTLS connections and their related DCI.¶
Either peer can add a new DTLS connection to the SCTP association at any time, but no more than 2 DTLS connections can exist at the same time. The new DCI value shall be the last active DCI increased by one modulo 4, this makes the attempt to create a new DTLS connection to use the same, known, value of DCI from both peers. A new handshake will be initiated by DTLS using the new DCI. Details of the handshake are described in Section 6.1.¶
As either endpoint can initiate a DTLS handshake at the same time, either endpoint may receive a DTLS ClientHello message when it has sent its own ClientHello. In this case the ClientHello from the endpoint that had the DTLS Client role in the establishment of the previous DTLS connection shall be continued to be processed and the other dropped.¶
When the handshake has been completed successfully, the new DTLS connection will be possible to use for traffic, if the handshake is not completed successfully, the new DCI value will not be considered used and a next attempt will reuse that DCI.¶
Either peers can initialize the removal of a DTLS connection from the current SCTP association when it is no longer the active one, i.e. when a newer DTLS connection is in use. It is RECOMMENDED to not initiate removal until at least one SCTP packet protected by the new DTLS connection has been received, and any transmitted packets protected using the new DTLS connection has been acknowledge, alternatively one Maximum Segment Lifetime (120 seconds) has passed since the last SCTP packet protected by the old DTLS connection was transmitted.¶
The closing of the DTLS connection when the SCTP association is in PROTECTED and ESTABLISHED state is done by having the DTLS connection send a DTLS close_notify. Note the difference in process for DTLS 1.2 and DTLS 1.3. Where sending the DTLS 1.2 close_notify will trigger an immediate close also in the peer. Which is why it is recommended to ensure that one have received packets from the peer using the new DTLS connection.¶
When DTLS closure for a DTLS connection is completed, the related DCI is released in the DTLS protection engine.¶
As DTLS has its own error reporting mechanism by exchanging DTLS alert messages no new DTLS related cause codes are defined to use the error handling defined in [I-D.westerlund-tsvwg-sctp-crypto-chunk].¶
When DTLS encounters an error it may report that issue using DTLS alert message to its peer by putting the created DTLS record in a DATA chunk with Protection Engine PPID and sending it in an SCTP packet. This is independent of what to do in relation to the SCTP association. Depending on the severance of the error different paths can be the result:¶
the DTLS connection can continue to protect the SCTP association. In this case the issue may be worth reporting to the peer using a DTLS alert message, but otherwise continue without further action.¶
If the DTLS connection has a critical issue, but can still protect packets then a the endpoint SHOULD attempt to establish a new DTLS connection. If that succeeds then the SCTP association switches over to the new DTLS connection and can terminate the old one including reporting the error. In case the establishment fails, then this critical issue MUST be reported to the SCTP association so that it can send an ABORT chunk with the Error in Protection cause code. This will terminate the SCTP association immediately, provide ULP with notification of the failure and speeding up any higher layer management of the failure.¶
If the DTLS connection fails so that no further data can be protected (i.e. either sent or received) with maintained security then it is not possible to establish a new DTLS connection and the protection engine will have to indicate this to the SCTP implementation so it can perform a one sides SCTP association termination. This will lead to an eventual SCTP association timeout in the peer.¶
This document defines the usage of either DTLS 1.3 [RFC9147], or DTLS 1.2 [RFC6347]. Earlier versions of DTLS MUST NOT be used (see [RFC8996]). DTLS 1.3 is RECOMMENDED for security and performance reasons. It is expected that DTLS in SCTP as described in this document will work with future versions of DTLS.¶
Only one version of DTLS MUST be used during the lifetime of an SCTP Association, meaning that the procedure for replacing the DTLS version in use requires the existing SCTP Association to be terminated and a new SCTP Association with the desired DTLS version to be instantiated.¶
The DTLS Connection ID SHALL NOT be included in the DTLS records as it is not needed, the CRYPTO chunk indicates which DTLS connection the DTLS records are intended for using the DCI bits. Avoiding overhead and addition implementation requirements on DTLS implementation.¶
The DTLS record length field is normally not needed as the CRYPTO Chunk provides a length field unless multiple records are put in same chunk payload. If multiple DTLS records are included in one CRYPTO chunk payload the DTLS record length field MUST be present in all but the last.¶
DTLS record replay detection MUST be used.¶
Sequence number size can be adapted based on how quickly it wraps.¶
Many of the TLS registries have a "Recommended" column. Parameters not marked as "Y" are NOT RECOMMENDED to support in DTLS in SCTP. Non-AEAD cipher suites or cipher suites without confidentiality MUST NOT be supported. Cipher suites and parameters that do not provide ephemeral key exchange MUST NOT be supported.¶
DTLS in SCTP MUST be mutually authenticated. Authentication is the process of establishing the identity of a user or system and verifying that the identity is valid. DTLS only provides proof of possession of a key. DTLS in SCTP MUST perform identity authentication. It is RECOMMENDED that DTLS in SCTP is used with certificate-based authentication. When certificates are used the application using DTLS in SCTP is responsible for certificate policies, certificate chain validation, and identity authentication (HTTPS does for example match the hostname with a subjectAltName of type dNSName). The application using DTLS in SCTP defines what the identity is and how it is encoded and the client and server MUST use the same identity format. Guidance on server certificate validation can be found in [I-D.ietf-uta-rfc6125bis]. DTLS in SCTP enables periodic transfer of mutual revocation information (OSCP stapling) every time a new parallel connection is set up. All security decisions MUST be based on the peer's authenticated identity, not on its transport layer identity.¶
It is possible to authenticate DTLS endpoints based on IP addresses in certificates. SCTP associations can use multiple IP addresses per SCTP endpoint. Therefore, it is possible that DTLS records will be sent from a different source IP address or to a different destination IP address than that originally authenticated. This is not a problem provided that no security decisions are made based on the source or destination IP addresses.¶
Implementations MUST set up new DTLS connections before any of the certificates expire. It is RECOMMENDED that all negotiated and exchanged parameters are the same except for the timestamps in the certificates. Clients and servers MUST NOT accept a change of identity during the setup of a new connections, but MAY accept negotiation of stronger algorithms and security parameters, which might be motivated by new attacks.¶
Allowing new connections can enable denial-of-service attacks. The endpoints MUST limit the number of simultaneous connections to two.¶
To force attackers to do dynamic key exfiltration and limits the amount of compromised data due to key compromise implementations MUST have policies for how often to set up new connections with ephemeral key exchange such as ECDHE. Implementations SHOULD set up new connections frequently to force attackers to dynamic key extraction. E.g., at least every hour and every 100 GB of data which is a common policy for IPsec [ANSSI-DAT-NT-003]. See [I-D.ietf-tls-rfc8446bis] for a more detailed discussion on key compromise and key exfiltration in (D)TLS.¶
For many DTLS in SCTP deployments the SCTP association is expected to have a very long lifetime of months or even years. For associations with such long lifetimes there is a need to frequently re-authenticate both client and server by setting up new connections. TLS Certificate lifetimes significantly shorter than a year are common which is shorter than many expected SCTP associations protected by DTLS in SCTP.¶
Both SCTP and DTLS contains mechanisms to padd SCTP payloads, and DTLS records respectively. If padding of SCTP packets are desired to hide actual message sizes it RECOMMEDED to use the SCTP Padding Chunck [RFC4820] to generate a consisted SCTP payload size. Support of this chunk is only required on the sender side. However, if the PAD chunk is not supported DTLS padding MAY be used.¶
It needs to be noted that independent if SCTP padding or DTLS padding is used the padding is not taken into account by the SCTP congestion control. Extensive use of padding has potential for worsen congestion situations as the SCTP association will consume more bandwidth than its derived share by the congestion control.¶
The use of SCTP PAD chunk is recommened as it at least can enable future extension or SCTP implementation that account also for the padding. Use of DTLS padding hides this packet expansion from SCTP.¶
The updates in Section 13 of [RFC9147] SHALL be followed for DTLS 1.2. DTLS 1.2 MUST be configured to disable options known to provide insufficient security. HTTP/2 [RFC9113] gives good minimum requirements based on the attacks that where publicly known in 2022.¶
The AEAD limits in DTLS 1.3 are equally valid for DTLS 1.2 and SHOULD be followed for DTLS in SCTP, but are not mandated by the DTLS 1.2 specification.¶
Use of renegotiation is NOT RECOMMENDED as it is disables in many implementations and does not provide any benefits in DTLS in SCTP compared to setting up a new connection. Resumption MAY be used but does not provide ephemeral key exchange as in DTLS 1.3¶
DTLS 1.3 is preferred over DTLS 1.2 being a newer protocol that addresses known vulnerabilities and only defines strong algorithms without known major weaknesses at the time of publication.¶
DTLS 1.3 requires rekeying before algorithm specific AEAD limits have been reached. Implementations MAY setup a new DTLS connection instead of using key update.¶
In DTLS 1.3 any number of tickets can be issued in a connection and the tickets can be used for resumption as long as they are valid, which is up to seven days. The nodes in a resumed connection have the same roles (client or server) as in the connection where the ticket was issued. Resumption can have significant latency benefits for quickly restarting a broken DTLS/SCTP association. If tickets and resumption are used it is enough to issue a single ticket per connection.¶
The PSK key exchange mode psk_ke MUST NOT be used as it does not provide ephemeral key exchange.¶
This section specifies how DTLS in SCTP is established after Protected Association Parameter with DTLS 1.2 or DTLS 1.3 as protection engine has been negotiated in the Init and Init-ACK exchange per [I-D.westerlund-tsvwg-sctp-crypto-chunk].¶
As soon the SCTP Association has entered the SCTP state PROTECTION PENDING as defined by [I-D.westerlund-tsvwg-sctp-crypto-chunk] the DTLS handshake procedure is initiated by the endpoint that has initiated the SCTP association.¶
The DTLS endpoint will if necessary fragment the handshake into multiple records each meeting the known or set MTU limit of the path between SCTP endpoints. Each DTLS handshake message fragment is sent as a SCTP user message on the same stream where each message is configured for reliable and in-order delivery with the Protection Engine PPID. The DTLS instance SHOULD NOT use DTLS retransmission to repair any packet losses of handshake message fragment. Note: If the DTLS implementation support configuring a MTU larger than the actual IP MTU it could be used as SCTP provides reliability and fragmentation.¶
If the DTLS handshake is successful in establishing a security context to protect further communication and the peer identity is accepted then the SCTP association is informed that it can move to the PROTECTED state.¶
If the DTLS handshake failed the SCTP association SHALL be aborted and an ERROR chunk with the Error in Protection error cause, with the appropriate extra error causes is generated, the right selection of "Error During Protection Handshake" or "Timeout During Protection Handshake or Validation".¶
When the SCTP Association has entered the ESTABLISHED state, each of the endpoint can initiate an DTLS handshake.¶
The DTLS endpoint will if necessary fragment the handshake into multiple records each meeting the known or set MTU limit of the path between SCTP endpoints. Each DTLS handshake message fragment is sent as a SCTP user message on the same stream where each message is configured for reliable and in-order delivery with the Protection Engine PPID. The DTLS instance SHOULD NOT use DTLS retransmission to repair any packet losses of handshake message fragment. Note: If the DTLS implementation support configuring a MTU larger than the actual IP MTU it could be used as SCTP provides reliability and fragmentation.¶
If the DTLS handshake failed the SCTP association SHALL generate an ERROR chunk with the Error in Protection error cause, with extra error causes "Error During Protection Handshake".¶
When the SCTP association has entered the PROTECTED state after the DTLS handshake has completed, the protection against downgrade in the negotiation of protection engine is performed per [I-D.westerlund-tsvwg-sctp-crypto-chunk]. The PVALID chunk will sent as a DTLS protected CRYPTO chunk payload per Section 7, thus protecting the plain text chunk.¶
If the validation completes successful the SCTP association will enter ESTABLISHED state. ULP data exchanges can now happen and will be protected together will all other SCTP packets.¶
CRYPTO chunk sending happens when SCTP requires transferring control or DATA chunk(s) to the remote SCTP Endpoint. For a proper handling, DCI shall be set to an established instance of DTLS connection.¶
SCTP Chunk handler will create the payload of a legacy SCTP packet according to [RFC9260] and any used SCTP extensions. Such payload will assume a PMTU that is equal to the value computed by SCTP minus the size of the CRYPTO Chunk header and DTLS record and authentication tag overhead. It's up to SCTP Chunk Handler to implement all the SCTP rules for bundling and retransmission mechanism. Once ready, the payload will be transferred to DTLS as a single array of bytes.¶
Once DTLS has created the related DTLS record (or DTLS records), it will transfer the encrypted data as an array of bytes to CRYPTO chunk handler for encapsulation into a CRYPTO chunk and being forwarded to the SCTP header handler for transmission.¶
The interface between SCTP and DTLS related to SCTP Payload will need to carefully evaluate the PMTU as seen by SCTP and DTLS so that each payload generated by SCTP Chunk Handler will not cause the finished SCTP packet to exceed the known path MTU unless it is a Path MTUD discovery packet.¶
When receiving an SCTP packet containing a CRYPTO Chunk it will contain an payload of protected SCTP control or data chunks. Since there's at most one CRYPTO Chunk per SCTP packet, the payload of that chunk will be transferred to the proper DTLS instance according to DCI for decryption and processing.¶
As discussed in CRYPTO Chunk specification when receiving packets certain meta data will be needed to associate with the protected CRYPTO chunk payload for SCTP to correctly process it. This includes packet size, source IP and arrival interface, i.e. path information, and ECN bits.¶
When DTLS processes a DTLS record with decryption and integrity verification and that contains application data, it will output the data as an array of bytes and transfer it back to the CRYPTO Handler that delivers it for SCTP chunk handling.¶
SCTP Chunk handler will threat the array as the payload of an SCTP packet, thus it will extract all the chunks and handle them according to [RFC9260] and any supported extension.¶
Rekeying in this specification is implemented by replacing the DTLS connection getting old with a new one. This feature exploits the capability of parallel DTLS connections and the possibility to add and remove DTLS connections during the lifetime of the SCTP Association.¶
The criteria for rekeying may vary depending on the ULP requirement on security properties, chosen cipher suits etc. Therefore it is assumed that the implementation will be configurable by the ULP to meet its demand.¶
Likely criteria to impact the need for rekeying through the usage of new DTLS connection are:¶
This specification allows up to 2 DTLS connection to be active at the same time for the current SCTP Association. The following state machine applies.¶
Trigger for rekeying can either be a local AGING event, triggered by the DTLS connection meeting the criteria for rekeying, or a REMOTE AGING event, triggered by receiving a DTLS record on the DCI that would be used for new DTLS connection. In such case a new DTLS connection shall be added according to Section 4.2.1 with a new DCI.¶
As soon as the new DTLS connection completes handshaking, the traffic is moved from the old one, then the procedure for closing the old DTLS connection is initiated, see Section 4.2.2.¶
A race condition may happen when both peer experience local AGING event at the same time and start creation of a new DTLS connection.¶
Since the criteria for calculating a new DCI is known and specified in Section 4.2.1, the peers will use the same DCI for identifying the new DTLS connection. And the race condition is solved as specified in Section 4.2.1.¶
Due to the DTLS record limitation for application data SCTP MUST use 214 as input to determine absolute maximum MTU when running PMTUD and using DTLS in SCTP as protection engine.¶
The DTLS protection engine MUST provide its maximum overhead for DTLS records and authentication tags when protecting the SCTP payload. This so that SCTP PMTUD can take this into consideration and ensure that produced packets that are not PMTUD probes does not become oversized. This may require updating during the SCTP associations lifetime due to future handshakes affecting cipher suit in use, or changes to record layer configurations.¶
Note that this implies that DTLS protection engine is expected to accept application data payloads of potentially larger sizes than what it configured to use for messages the DTLS implementation generates itself for signaling.¶
The security considerations given in [RFC9147], [RFC6347], and [RFC9260] also apply to this document. BCP 195 [RFC9325] [RFC8996] provides recommendations and requirements for improving the security of deployed services that use DTLS. BCP 195 MUST be followed which implies that DTLS 1.0 SHALL NOT be supported and are therefore not defined.¶
Although DTLS in SCTP provides privacy for the actual user message as well as almost all chunks, some fields are not confidentiality protected. In addition to the DTLS record header, the SCTP common header and the CRYPTO chunk header are not confidentiality protected. An attacker can correlate DTLS connections over the same SCTP association using the SCTP common header.¶
To provide identity protection it is RECOMMENDED that DTLS in SCTP is used with certificate-based authentication in DTLS 1.3 [RFC9147] and to not reuse tickets. DTLS 1.2 and DTLS 1.3 with external PSK authentication does not provide identity protection.¶
By mandating ephemeral key exchange and cipher suites with confidentiality DTLS in SCTP effectively mitigate many forms of passive pervasive monitoring. By recommending implementations to frequently set up new DTLS connections with (EC)DHE force attackers to do dynamic key exfiltration and limits the amount of compromised data due to key compromise.¶
This document adds the two new entries listed in Table 1 into the "CRYPTO Chunk Protection Engine Identifiers" registry in the Stream Control Transmission Protocol (SCTP) Parameters grouping.¶
IANA is requested to register two Protection Engine Identifiers in the "CRYPTO Chunk Protection Engine Identifiers" registry defined by [I-D.westerlund-tsvwg-sctp-crypto-chunk]. The entries to be registered are provided in Table 2.¶
ID VALUE | Name | Reference | Contact |
---|---|---|---|
0 | DTLS 1.2 | RFC-To-Be | Authors |
1 | DTLS 1.3 | RFC-To-Be | Authors |