| Internet-Draft | HTTP UDP CONNECT | March 2022 |
| Schinazi | Expires 6 September 2022 | [Page] |
This document describes how to proxy UDP over HTTP. Similar to how the CONNECT method allows proxying TCP over HTTP, this document defines a new mechanism to proxy UDP. It is built using HTTP Extended CONNECT.¶
This note is to be removed before publishing as an RFC.¶
Discussion of this document takes place on the MASQUE WG mailing list (masque@ietf.org), which is archived at https://mailarchive.ietf.org/arch/browse/masque/.¶
Source for this draft and an issue tracker can be found at https://github.com/ietf-wg-masque/draft-ietf-masque-connect-udp.¶
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 6 September 2022.¶
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 how to proxy UDP over HTTP. Similar to how the CONNECT method (see Section 9.3.6 of [HTTP]) allows proxying TCP [TCP] over HTTP, this document defines a new mechanism to proxy UDP [UDP].¶
UDP Proxying supports all versions of HTTP and uses HTTP Datagrams [HTTP-DGRAM]. When using HTTP/2 or HTTP/3, UDP proxying uses HTTP Extended CONNECT as described in [EXT-CONNECT2] and [EXT-CONNECT3]. When using HTTP/1.x, UDP proxying uses HTTP Upgrade as defined in Section 7.8 of [HTTP].¶
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.¶
In this document, we use the term "proxy" to refer to the HTTP server that opens the UDP socket and responds to the UDP proxying request. If there are HTTP intermediaries (as defined in Section 3.7 of [HTTP]) between the client and the proxy, those are referred to as "intermediaries" in this document.¶
Note that, when the HTTP version in use does not support multiplexing streams (such as HTTP/1.1), any reference to "stream" in this document represents the entire connection.¶
Clients are configured to use UDP Proxying over HTTP via an URI Template [TEMPLATE]. The URI template MUST contain exactly two variables: "target_host" and "target_port". Examples are shown below:¶
https://masque.example.org/{target_host}/{target_port}/
https://proxy.example.org:4443/masque?h={target_host}&p={target_port}
https://proxy.example.org:4443/masque{?target_host,target_port}
Since the original HTTP CONNECT method allowed conveying the target host and port but not the scheme, proxy authority, path, nor query, there exist proxy configuration interfaces that only allow the user to configure the proxy host and the proxy port. Client implementations of this specification that are constrained by such limitations MUST use the default template which is defined as: "https://$PROXY_HOST:$PROXY_PORT/{target_host}/{target_port}/" where $PROXY_HOST and $PROXY_PORT are the configured host and port of the proxy respectively. Proxy deployments SHOULD use the default template to facilitate interoperability with such clients.¶
This document defines the "connect-udp" HTTP Upgrade Token. "connect-udp" uses the Capsule Protocol as defined in [HTTP-DGRAM].¶
A "connect-udp" request requests that the recipient proxy establish a tunnel over a single HTTP stream to the destination target identified by the "target_host" and "target_port" variables of the URI template (see Section 2). If the request is successful, the proxy commits to converting received HTTP Datagrams into UDP packets and vice versa until the tunnel is closed. Tunnels are commonly used to create an end-to-end virtual connection, which can then be secured using QUIC [QUIC] or another protocol running over UDP.¶
When sending its UDP proxying request, the client SHALL perform URI template expansion to determine the path and query of its request. target_host supports using DNS names, IPv6 literals and IPv4 literals. Note that this URI template expansion requires using pct-encoding, so for example if the target_host is "2001:db8::42", it will be encoded in the URI as "2001%3Adb8%3A%3A42".¶
A payload within a UDP proxying request message has no defined semantics; a UDP proxying request with a non-empty payload is malformed.¶
Responses to UDP proxying requests are not cacheable.¶
Upon receiving a UDP proxying request, the recipient proxy extracts the "target_host" and "target_port" variables from the URI it has reconstructed from the request headers, and establishes a tunnel by directly opening a UDP socket to the requested target.¶
Unlike TCP, UDP is connection-less. The proxy that opens the UDP socket has no way of knowing whether the destination is reachable. Therefore it needs to respond to the request without waiting for a packet from the target. However, if the target_host is a DNS name, the proxy MUST perform DNS resolution before replying to the HTTP request. If DNS resolution fails, the proxy MUST fail the request and SHOULD send details using the Proxy-Status header [PROXY-STATUS].¶
Proxies can use connected UDP sockets if their operating system supports them, as that allows the proxy to rely on the kernel to only send it UDP packets that match the correct 5-tuple. If the proxy uses a non-connected socket, it MUST validate the IP source address and UDP source port on received packets to ensure they match the client's request. Packets that do not match MUST be discarded by the proxy.¶
The lifetime of the socket is tied to the request stream. The proxy MUST keep the socket open while the request stream is open. If a proxy is notified by its operating system that its socket is no longer usable (for example, this can happen when an ICMP "Destination Unreachable" message is received, see Section 3.1 of [ICMP6]), it MUST close the request stream. Proxies MAY choose to close sockets due to a period of inactivity, but they MUST close the request stream when closing the socket. Proxies that close sockets after a period of inactivity SHOULD NOT use a period lower than two minutes, see Section 4.3 of [BEHAVE].¶
A successful response (as defined in Section 3.3 and Section 3.5) indicates that the proxy has opened a socket to the requested target and is willing to proxy UDP payloads. Any response other than a successful response indicates that the request has failed, and the client MUST therefore abort the request.¶
Proxies MUST NOT introduce fragmentation at the IP layer when forwarding HTTP Datagrams onto a UDP socket. In IPv4, the Don't Fragment (DF) bit MUST be set if possible, to prevent fragmentation on the path. Future extensions MAY remove these requirements.¶
When using HTTP/1.1, a UDP proxying request will meet the following requirements:¶
For example, if the client is configured with URI template "https://proxy.example.org/{target_host}/{target_port}/" and wishes to open a UDP proxying tunnel to target 192.0.2.42:443, it could send the following request:¶
CONNECT https://proxy.example.org/192.0.2.42/443/ HTTP/1.1 Host: proxy.example.org Connection: upgrade Upgrade: connect-udp
The proxy SHALL indicate a successful response by replying with the following requirements:¶
If any of these requirements are not met, the client MUST treat this proxying attempt as failed and abort the connection.¶
For example, the proxy could respond with:¶
HTTP/1.1 101 Switching Protocols Connection: upgrade Upgrade: connect-udp
When using HTTP/2 [H2] or HTTP/3 [H3], UDP proxying requests use HTTP pseudo-headers with the following requirements:¶
A UDP proxying request that does not conform to these restrictions is malformed (see Section 8.1.1 of [H2]).¶
For example, if the client is configured with URI template "https://proxy.example.org/{target_host}/{target_port}/" and wishes to open a UDP proxying tunnel to target 192.0.2.42:443, it could send the following request:¶
HEADERS :method = CONNECT :protocol = connect-udp :scheme = https :path = /192.0.2.42/443/ :authority = proxy.example.org
The proxy SHALL indicate a successful response by replying with any 2xx (Successful) HTTP status code, without any Transfer-Encoding or Content-Length header fields.¶
If any of these requirements are not met, the client MUST treat this proxying attempt as failed and abort the request.¶
For example, the proxy could respond with:¶
HEADERS :status = 200
[[RFC editor: please remove this section before publication.]]¶
In order to allow implementations to support multiple draft versions of this specification during its development, we introduce the "connect-udp-version" header. When sent by the client, it contains a list of draft numbers supported by the client (e.g., "connect-udp-version: 0, 2"). When sent by the proxy, it contains a single draft number selected by the proxy from the list provided by the client (e.g., "connect-udp-version: 2"). Sending this header is RECOMMENDED but not required. Its ABNF is:¶
connect-udp-version = sf-list¶
This protocol allows future extensions to exchange HTTP Datagrams which carry different semantics from UDP payloads. Some of these extensions can augment UDP payloads with additional data, while others can exchange data that is completely separate from UDP payloads. In order to accomplish this, all HTTP Datagrams associated with UDP Proxying request streams start with a context ID, see Section 5.¶
Context IDs are 62-bit integers (0 to 262-1). Context IDs are encoded as variable-length integers, see Section 16 of [QUIC]. The context ID value of 0 is reserved for UDP payloads, while non-zero values are dynamically allocated: non-zero even-numbered context IDs are client-allocated, and odd-numbered context IDs are proxy-allocated. The context ID namespace is tied to a given HTTP request: it is possible for a context ID with the same numeric value to be simultaneously assigned different semantics in distinct requests, potentially with different semantics. Context IDs MUST NOT be re-allocated within a given HTTP namespace but MAY be allocated in any order. Once allocated, any context ID can be used by both client and proxy - only allocation carries separate namespaces to avoid requiring synchronization.¶
Registration is the action by which an endpoint informs its peer of the semantics and format of a given context ID. This document does not define how registration occurs, though some examples of how it might occur are provided in Appendix A. Depending on the method being used, it is possible for datagrams to be received with Context IDs which have not yet been registered, for instance due to reordering of the datagram and the registration packets during transmission.¶
When associated with UDP proxying request streams, the HTTP Datagram Payload field of HTTP Datagrams (see [HTTP-DGRAM]) has the format defined in Figure 6. Note that when HTTP Datagrams are encoded using QUIC DATAGRAM frames, the Context ID field defined below directly follows the Quarter Stream ID field which is at the start of the QUIC DATAGRAM frame payload:¶
UDP Proxying HTTP Datagram Payload {
Context ID (i),
Payload (..),
}
A variable-length integer that contains the value of the Context ID. If an HTTP/3 datagram which carries an unknown Context ID is received, the receiver SHALL either drop that datagram silently or buffer it temporarily (on the order of a round trip) while awaiting the registration of the corresponding Context ID.¶
The payload of the datagram, whose semantics depend on value of the previous field. Note that this field can be empty.¶
UDP packets are encoded using HTTP Datagrams with the Context ID set to zero. When the Context ID is set to zero, the Payload field contains the unmodified payload of a UDP packet (referred to as "data octets" in [UDP]).¶
Clients MAY optimistically start sending proxied UDP packets before receiving the response to its UDP proxying request, noting however that those may not be processed by the proxy if it responds to the request with a failure, or if the datagrams are received by the proxy before the request.¶
Endpoints MUST NOT send HTTP Datagrams with payloads longer than 65527 using Context ID zero. An endpoint that receives a DATAGRAM capsule using Context ID zero whose payload is longer than 65527 MUST abort the stream. If a proxy knows it can only send out UDP packets of a certain length due to its underlying link MTU, it SHOULD discard incoming DATAGRAM capsules using Context ID zero whose payload is longer than that limit without buffering the capsule contents.¶
Proxies SHOULD strive to avoid increasing burstiness of UDP traffic: they SHOULD NOT queue packets in order to increase batching.¶
When the protocol running over UDP that is being proxied uses congestion control (e.g., [QUIC]), the proxied traffic will incur at least two nested congestion controllers. This can reduce performance but the underlying HTTP connection MUST NOT disable congestion control unless it has an out-of-band way of knowing with absolute certainty that the inner traffic is congestion-controlled.¶
If a client or proxy with a connection containing a UDP proxying request stream disables congestion control, it MUST NOT signal ECN support on that connection. That is, it MUST mark all IP headers with the Not-ECT codepoint. It MAY continue to report ECN feedback via ACK_ECN frames, as the peer may not have disabled congestion control.¶
When the protocol running over UDP that is being proxied uses loss recovery (e.g., [QUIC]), and the underlying HTTP connection runs over TCP, the proxied traffic will incur at least two nested loss recovery mechanisms. This can reduce performance as both can sometimes independently retransmit the same data. To avoid this, UDP proxying SHOULD be performed over HTTP/3 to allow leveraging the QUIC DATAGRAM frame.¶
When using HTTP/3 with the QUIC Datagram extension [DGRAM], UDP payloads are transmitted in QUIC DATAGRAM frames. Since those cannot be fragmented, they can only carry payloads up to a given length determined by the QUIC connection configuration and the path MTU. If a proxy is using QUIC DATAGRAM frames and it receives a UDP payload from the target that will not fit inside a QUIC DATAGRAM frame, the proxy SHOULD NOT send the UDP payload in a DATAGRAM capsule, as that defeats the end-to-end unreliability characteristic that methods such as Datagram Packetization Layer Path MTU Discovery (DPLPMTUD) depend on [DPLPMTUD]. In this scenario, the proxy SHOULD drop the UDP payload and send an ICMP "Packet Too Big" message to the target, see Section 3.2 of [ICMP6].¶
UDP proxying does not create an IP-in-IP tunnel, so the guidance in [ECN-TUNNEL] about transferring ECN marks between inner and outer IP headers does not apply. There is no inner IP header in UDP proxying tunnels.¶
Note that UDP proxying clients do not have the ability in this specification to control the ECN codepoints on UDP packets the proxy sends to the target, nor can proxies communicate the markings of each UDP packet from target to proxy.¶
A UDP proxy MUST ignore ECN bits in the IP header of UDP packets received from the target, and MUST set the ECN bits to Not-ECT on UDP packets it sends to the target. These do not relate to the ECN markings of packets sent between client and proxy in any way.¶
There are significant risks in allowing arbitrary clients to establish a tunnel to arbitrary targets, as that could allow bad actors to send traffic and have it attributed to the proxy. Proxies that support UDP proxying SHOULD restrict its use to authenticated users.¶
Because the CONNECT method creates a TCP connection to the target, the target has to indicate its willingness to accept TCP connections by responding with a TCP SYN-ACK before the proxy can send it application data. UDP doesn't have this property, so a UDP proxy could send more data to an unwilling target than a CONNECT proxy. However, in practice denial of service attacks target open TCP ports so the TCP SYN-ACK does not offer much protection in real scenarios.¶
This document will request IANA to register "connect-udp" in the HTTP Upgrade Token Registry maintained at <https://www.iana.org/assignments/http-upgrade-tokens>.¶
Extensions can define new semantics for the payload of HTTP Datagrams. The extension can then have an endpoint pick an available locally-allocated context ID (see Section 4) and register that context ID with their peer.¶
Note that this appendix only exists to help illustrate MASQUE Working Group discussions while designing extensions. This appendix will be removed before MASQUE Working Group Last Call.¶
Extensions can define a new HTTP header to register a context ID with the peer endpoint.¶
As an example, take an extension that conveys the time at which a UDP packet was received. The extension would first define the format of its HTTP Datagram Payload field:¶
UDP with Timestamp HTTP Datagrams {
Context ID (i),
Timestamp (64),
UDP Payload (..),
}
The extension would also define a new HTTP header (Example-UDP-Timestamps) that includes a context ID value. Proxies that understand this new HTTP header would be able to consequently handle and parse datagrams with the context ID, while all other proxies would silently drop the datagrams.¶
This specific extension would restrict registrations to the client, and have them be bidirectional in the sense that the client registering a context ID also indicates support for receiving on it. Other extensions could allow proxy registrations, and/or unidirectional registrations in the sense that registration would only imply usage in one direction.¶
HEADERS :method = CONNECT :protocol = connect-udp :scheme = https :path = /192.0.2.42/443/ :authority = proxy.example.org example-udp-timestamps = 42
In this example request, HTTP Datagrams with context ID zero would only contain the UDP payload, whereas HTTP Datagrams with context ID 42 would also contain a timestamp.¶
Extensions can define a new Capsule type (see [HTTP-DGRAM]) to register a context ID with the peer endpoint.¶
As an example, take an extension that compresses QUIC Connection IDs when the client is running QUIC over a UDP proxying tunnel. The extension would first define the transform applied to UDP payloads when compressing and decompressing, such as removing the bytes of the connection ID.¶
The extension would also define a new capsule type (EXAMPLE_REGISTER_COMPRESSED_QUIC_CID) that includes a context ID value and the connection ID to compress. Endpoints that understand this new capsule type would be able to consequently handle and parse datagrams on the context ID, while all other endpoints would ignore the datagrams.¶
EXAMPLE_REGISTER_COMPRESSED_QUIC_CID Capsule {
Type (i) = EXAMPLE_REGISTER_COMPRESSED_QUIC_CID,
Length (i),
Context ID (i),
QUIC Connection ID (..),
}
This example extension would most likely also define a new HTTP header to indicate support.¶
This document is a product of the MASQUE Working Group, and the author thanks all MASQUE enthusiasts for their contibutions. This proposal was inspired directly or indirectly by prior work from many people. In particular, the author would like to thank Eric Rescorla for suggesting to use an HTTP method to proxy UDP. Thanks to Lucas Pardue for their inputs on this document. The extensibility design in this document came out of the HTTP Datagrams Design Team, whose members were Alan Frindell, Alex Chernyakhovsky, Ben Schwartz, Eric Rescorla, Lucas Pardue, Marcus Ihlar, Martin Thomson, Mike Bishop, Tommy Pauly, Victor Vasiliev, and the author of this document.¶