| Internet-Draft | UDP Encapsulation of SCTP Packets | August 2022 |
| Tuexen & Stewart | Expires 26 February 2023 | [Page] |
This document describes a simple method of encapsulating Stream Control Transmission Protocol (SCTP) packets into UDP packets and its limitations. This allows the usage of SCTP in networks with legacy NATs that do not support SCTP. It can also be used to implement SCTP on hosts without directly accessing the IP layer, for example, implementing it as part of the application without requiring special privileges.¶
Please note that this document only describes the functionality required within an SCTP stack to add on UDP encapsulation, providing only those mechanisms for two end-hosts to communicate with each other over UDP ports. In particular, it does not provide mechanisms to determine whether UDP encapsulation is being used by the peer, nor the mechanisms for determining which remote UDP port number can be used. These functions are out of scope for this document.¶
This document covers only end-hosts and not tunneling (egress or ingress) endpoints.¶
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 26 February 2023.¶
Copyright (c) 2022 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 a simple method of encapsulating SCTP packets into UDP packets. SCTP, as defined in [RFC4960], runs directly over IPv4 or IPv6. There are two main reasons for encapsulating SCTP packets:¶
SCTP provides the necessary congestion control and reliability service that UDP does not perform.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].¶
This section discusses two important use cases for encapsulating SCTP into UDP.¶
Some operating systems support SCTP natively. For other operating systems, implementations are available but require special privileges to install and/or use them. In some cases, a kernel implementation might not be available at all. When providing an SCTP implementation as part of a user process, most operating systems require special privileges to access the IP layer directly.¶
Using UDP encapsulation makes it possible to provide an SCTP implementation as part of a user process that does not require any special privileges.¶
A crucial point for implementing SCTP in user space is that the source address of outgoing packets needs to be controlled. This is not an issue if the SCTP stack can use all addresses configured at the IP layer as source addresses. However, it is an issue when also using the address management required for NAT traversal, described in Section 5.7.¶
Using UDP encapsulation allows SCTP communication when traversing legacy NATs (i.e, those NATs not supporting SCTP as described in [BEHAVE] and [NATSUPP]). For single-homed associations, IP addresses MUST NOT be listed in the INIT and INIT-ACK chunks. To use multiple addresses, the dynamic address reconfiguration extension described in [RFC5061] MUST be used only with wildcard addresses in the ASCONF chunks (Address Configuration Change Chunks) in combination with [RFC4895].¶
For multihomed SCTP associations, the address management as described in Section 5.7 MUST be performed.¶
SCTP sends periodic HEARTBEAT chunks on all idle paths. These can keep the NAT state alive.¶
As [RFC3424] requires a limited scope, this document only covers SCTP endpoints dealing with legacy constraints as described in Section 3. It doesn't cover generic tunneling endpoints.¶
Obviously, the exit strategy is to use hosts supporting SCTP natively and middleboxes supporting SCTP as specified in [BEHAVE] and [NATSUPP].¶
UDP-encapsulated SCTP is normally communicated between SCTP stacks using the IANA-assigned UDP port number 9899 (sctp-tunneling) on both ends. There are circumstances where other ports may be used on either end: As stated earlier, implementations in the application space might be required to use ports other than the registered port. Since NAT boxes might change UDP port numbers, the receiver might observe other UDP port numbers than were used by the sender. Discovery of alternate ports is outside of the scope of this document, but this section describes considerations for SCTP stack design in light of their potential use.¶
Each SCTP stack uses a single local UDP encapsulation port number as the destination port for all its incoming SCTP packets. While the uniqueness of the local UDP encapsulation port number is not necessarily required for the protocol, this greatly simplifies implementation design, since different ports for each address would require a sender implementation to choose the appropriate port while doing source address selection. Using a single local UDP encapsulation port number per host is not possible if the SCTP stack is implemented as part of each application, there are multiple applications, and some of the applications want to use the same IP address.¶
An SCTP implementation supporting UDP encapsulation MUST maintain a remote UDP encapsulation port number per destination address for each SCTP association. Again, because the remote stack may be using ports other than the well-known port, each port may be different from each stack. However, because of remapping of ports by NATs, the remote ports associated with different remote IP addresses may not be identical, even if they are associated with the same stack.¶
Implementation note: Because the well-known port might not be used, implementations need to allow other port numbers to be specified as a local or remote UDP encapsulation port number through APIs.¶
To encapsulate an SCTP packet, a UDP header as defined in [RFC0768] is inserted between the IP header as defined in [RFC0791] and the SCTP common header as defined in [RFC4960].¶
Figure 1 shows the packet format of an encapsulated SCTP packet when IPv4 is used.¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | IPv4 Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | UDP Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SCTP Common Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SCTP Chunk #1 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SCTP Chunk #n | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The packet format for an encapsulated SCTP packet when using IPv6 as defined in [RFC2460] is shown in Figure 2. Please note that the number m of IPv6 extension headers can be 0.¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | IPv6 Base Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | IPv6 Extension Header #1 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | IPv6 Extension Header #m | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | UDP Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SCTP Common Header | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SCTP Chunk #1 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SCTP Chunk #n | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Within the UDP header, the source port MUST be the local UDP encapsulation port number of the SCTP stack, and the destination port MUST be the remote UDP encapsulation port number maintained for the association and the destination address to which the packet is sent (see Section 5.1).¶
Because the SCTP packet is the UDP payload, the length of the UDP packet MUST be the length of the SCTP packet plus the size of the UDP header.¶
The SCTP checksum MUST be computed for IPv4 and IPv6, and the UDP checksum SHOULD be computed for IPv4 and IPv6. (See [RFC0768] regarding IPv4; see [RFC2460] and [RFC6936] regarding IPv6.) Although UDP with a zero checksum over IPv6 is allowed under certain constraints [RFC6936], this document does not specify mechanisms for this mode. Deployed support may be limited; also, at the time of writing, the use of a zero UDP checksum would be counter to the goal of legacy NAT traversal.¶
When an encapsulated packet is received, the UDP header is removed. Then, the generic lookup is performed, as done by an SCTP stack whenever a packet is received, to find the association for the received SCTP packet. After finding the SCTP association (which includes checking the verification tag), the UDP source port MUST be stored as the encapsulation port for the destination address the SCTP packet is received from (see Section 5.1).¶
When a non-encapsulated SCTP packet is received by the SCTP stack, the encapsulation of outgoing packets belonging to the same association and the corresponding destination address MUST be disabled.¶
When receiving ICMP or ICMPv6 response packets, there might not be enough bytes in the payload to identify the SCTP association that the SCTP packet triggering the ICMP or ICMPv6 packet belongs to. If a received ICMP or ICMPv6 packet cannot be related to a specific SCTP association or the verification tag cannot be verified, it MUST be discarded silently. In particular, this means that the SCTP stack MUST NOT rely on receiving ICMP or ICMPv6 messages. Implementation constraints could prevent processing received ICMP or ICMPv6 messages.¶
If received ICMP or ICMPv6 messages are processed, the following mapping SHOULD apply:¶
If an SCTP endpoint starts to encapsulate the packets of a path, it MUST decrease the Path MTU of that path by the size of the UDP header. If it stops encapsulating them, the Path MTU SHOULD be increased by the size of the UDP header.¶
When performing Path MTU discovery as described in [RFC4820] and [RFC4821], it MUST be taken into account that one cannot rely on the feedback provided by ICMP or ICMPv6 due to the limitation laid out in Section 5.5.¶
If the implementation does not allow control of the Don't Fragment (DF) bit contained in the IPv4 header, then Path MTU discovery can't be used. In this case, an implementation-specific value should be used instead.¶
When using UDP encapsulation for legacy NAT traversal, IP addresses that might require translation MUST NOT be put into any SCTP packet.¶
This means that a multihomed SCTP association is set up initially as a single-homed one, and the protocol extension [RFC5061] in combination with [RFC4895] is used to add the other addresses. Only wildcard addresses are put into the SCTP packet.¶
When addresses are changed during the lifetime of an association, the protocol extension [RFC5061] MUST be used with wildcard addresses only. If an SCTP endpoint receives an ABORT with the T-bit set, it MAY use this as an indication that the addresses seen by the peer might have changed.¶
If the implementation supports the sending and receiving of the ECN bits for the IP protocols being used by an SCTP association, the ECN bits MUST NOT be changed during sending and receiving.¶
This section describes how the socket API defined in [RFC6458] needs to be extended to provide a way for the application to control the UDP encapsulation.¶
Please note that this section is informational only.¶
A socket API implementation based on [RFC6458] is extended by supporting one new read/write socket option.¶
This socket option can be used to set and retrieve the UDP encapsulation port number. This allows an endpoint to encapsulate initial packets.¶
struct sctp_udpencaps {
sctp_assoc_t sue_assoc_id;
struct sockaddr_storage sue_address;
uint16_t sue_port;
};
¶
This document refers to the already assigned UDP port 9899 (sctp-tunneling). IANA has updated this assignment to refer to this document. As per [RFC6335], the Assignee is [IESG] and the Contact is [IETF_Chair].¶
Please note that the TCP port 9899 (sctp-tunneling) assignment is not needed anymore, and IANA has removed this TCP port number assignment and marked TCP port 9899 as "Reserved".¶
Encapsulating SCTP into UDP does not add any additional security considerations to the ones given in [RFC4960] and [RFC5061].¶
Firewalls inspecting SCTP packets must also be aware of the encapsulation and apply corresponding rules to the encapsulated packets.¶
An attacker might send a malicious UDP packet towards an SCTP endpoint to change the encapsulation port for a single remote address of a particular SCTP association. However, as specified in Section 5.4, this requires the usage of one of the two negotiated verification tags. This protects against blind attackers the same way as described in [RFC4960] for SCTP over IPv4 or IPv6. Non-blind attackers can affect SCTP association using the UDP encapsulation described in this document in the same way as SCTP associations not using the UDP encapsulation of SCTP described here.¶
The authors wish to thank Stewart Bryant, Dave Crocker, Gorry Fairhurst, Tero Kivinen, Barry Leiba, Pete Resnick, Martin Stiemerling, Irene Ruengeler, and Dan Wing for their invaluable comments.¶