| TAPS Working Group | T. Pauly, Ed. |
| Internet-Draft | Apple Inc. |
| Intended status: Standards Track | B. Trammell, Ed. |
| Expires: April 25, 2019 | ETH Zurich |
| A. Brunstrom | |
| Karlstad University | |
| G. Fairhurst | |
| University of Aberdeen | |
| C. Perkins | |
| University of Glasgow | |
| P. Tiesel | |
| TU Berlin | |
| C. Wood | |
| Apple Inc. | |
| October 22, 2018 |
An Architecture for Transport Services
draft-ietf-taps-arch-02
This document provides an overview of the architecture of Transport Services, a system for exposing the features of transport protocols to applications. This architecture serves as a basis for Application Programming Interfaces (APIs) and implementations that provide flexible transport networking services. It defines the common set of terminology and concepts to be used in more detailed discussion of Transport Services.
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 April 25, 2019.
Copyright (c) 2018 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Many application programming interfaces (APIs) to perform transport networking have been deployed, perhaps the most widely known and imitated being the BSD socket() [POSIX] interface. The names and functions between these APIs are not consistent, and vary depending on the protocol being used. For example, sending and receiving on a stream of data is conceptually the same between operating on an unencrypted Transmission Control Protocol (TCP) stream and operating on an encrypted Transport Layer Security (TLS) [I-D.ietf-tls-tls13] stream over TCP, but applications cannot use the same socket send() and recv() calls on top of both kinds of connections. Similarly, terminology for the implementation of protocols offering transport services vary based on the context of the protocols themselves. This variety can lead to confusion when trying to understand the similarities and differences between protocols, and how applications can use them effectively.
The goal of the Transport Services architecture is to provide a common, flexible, and reusable interface for transport protocols. As applications adopt this interface, they will benefit from a wide set of transport features that can evolve over time, and ensure that the system providing the interface can optimize its behavior based on the application requirements and network conditions.
This document is developed in parallel with the specification of the Transport Services API [I-D.ietf-taps-interface] and Implementation [I-D.ietf-taps-impl] documents.
The model for using sockets for networking can be represented as follows: applications create connections and transfer data using the socket API, which provides the interface to the implementations of UDP and TCP (typically implemented in the system’s kernel), which in turn send data over the available network layer interfaces.
+-----------------------------------------------------+
| Application |
+-----------------------------------------------------+
| |
+---------------------+ +-----------------------+
| Socket Stream API | | Socket Datagram API |
+---------------------+ +-----------------------+
| |
+-----------------------------------------------------+
| TCP UDP |
| Kernel Protocol Implementation |
+-----------------------------------------------------+
|
+-----------------------------------------------------+
| Network Layer Interface |
+-----------------------------------------------------+
The Transport Services architecture maintains this general model of interaction, but aims to both modernize the API surface exposed for transport protocols and enrich the capabilities of the transport system implementation.
+-----------------------------------------------------+
| Application |
+-----------------------------------------------------+
|
+-----------------------------------------------------+
| Transport Services API |
+-----------------------------------------------------+
|
+-----------------------------------------------------+
| Transport System Implementation |
| (UDP, TCP, SCTP, DCCP, TLS, QUIC, etc) |
+-----------------------------------------------------+
|
+-----------------------------------------------------+
| Network Layer Interface |
+-----------------------------------------------------+
The Transport Services API [I-D.ietf-taps-interface] defines the mechanism for an application to create and monitor network connections, and transfer data. The Implementation [I-D.ietf-taps-impl] is responsible for mapping the API into the various available transport protocols and managing the available network interfaces and paths.
There are a few key departures that Transport Services makes from the sockets API: it presents an asynchronous, event-driven API; it uses messages for respresenting data transfer to applications; and it assumes an implementation that can use multiple IP addresses, multiple protocols, multiple paths, and provide multiple application streams.
Originally, sockets presented a blocking interface for establishing connections and transferring data. However, most modern applications interact with the network asynchronously. When sockets are presented as an asynchronous interface, they generally use a try-and-fail model. If the application wants to read, but data has not yet been received from the peer, the call to read will fail. The application then waits for a notification that it should try again.
All interaction with a Transport Services system is expected to be asynchronous, and use an event-driven model unlike sockets Section 4.1.5. For example, if the application wants to read, its call to read will not fail, but will deliver an event containing the received data once it is available.
The Transport Services API also delivers events regarding the lifetime of a connection and changes to available network links, which were not previously made explicit in sockets.
Using asynchronous events allows for a much simpler interaction model when establishing connections and transferring data. Events in time more closely reflect the nature of interactions over networks, as opposed to how sockets represent network resources as file system objects that may be temporarily unavailable.
Sockets provide a message interface for datagram protocols like UDP, but provide an unstructured stream abstraction for TCP. While TCP does indeed provide the ability to send and receive data as streams, most applications need to interpret structure within these streams. HTTP/1.1 uses character delimiters to segment messages over a stream; TLS record headers carry a version, content type, and length; and HTTP/2 uses frames to segment its headers and bodies.
In order to more closely match the way applications use the network, the Transport Services API respresents data as messages. Messages seamlessly work with transport protocols that support datagrams or records, but can also be used over a stream by defining the application-layer framing being used Section 4.4.
Sockets, for protocols like TCP, are generally limited to connecting to a single address over a single interface. They also present a single stream to the application. Software layers built upon sockets often propagate this limitation of a single-address single-stream model. The Transport Services architecture is designed to handle multiple candidate endpoints, protocols, and paths; and support multipath and multistreaming protocols.
Transport Services implementations are meant to be flexible at connection establishment time, considering many different options and trying to select the most optimal combinations (Section 4.2.1 and Section 4.2.2). This requires applications to provide higher-level endpoints than IP addresses, such as hostnames and URLs, which are used by a Transport Services implementation for resolution, path selection, and racing.
Flexibility after connection establishment is also important. Transport protocols that can migrate between multiple network layer interfaces need to be able to process and react to interface changes. Protocols that support multiple application-layer streams need to support initiating and receiving new streams using existing connections.
The Transport Services architecture is based on the survey of Services Provided by IETF Transport Protocols and Congestion Control Mechanisms [RFC8095], and the distilled minimal set of the features offered by transport protocols [I-D.ietf-taps-minset]. This work has identified common features and patterns across all transport protocols developed thus far in the IETF.
Since transport security is an increasingly relevant aspect of using transport protocols on the Internet, this architecture also considers the impact of transport security protocols on the feature set exposed by transport services [I-D.ietf-taps-transport-security].
One of the key insights to come from identifying the minimal set of features provided by transport protocols [I-D.ietf-taps-minset] was that features either require application interaction and guidance (referred to as Functional Features), or else can be handled automatically by a system implementing Transport Services (referred to as Automatable Features). Among the Functional Features, some were common across all or nearly all transport protocols, while others could be seen as features that, if specified, would only be useful with a subset of protocols, or perhaps even a single transport protocol, but would not harm the functionality of other protocols. For example, some protocols can deliver messages faster for applications that do not require them to arrive in the order in which they were sent. However, this functionality must be explicitly allowed by the application, since reordering messages would be undesirable in many cases.
The goal of the Transport Services architecture is to redefine the interface between applications and transports in a way that allows the transport layer to evolve and improve without fundamentally changing the contract with the application. This requires a careful consideration of how to expose the capabilities of protocols.
There are several degrees in which a Transport Services system can offer flexibility to an application: it can provide access to multiple sets of protocols and protocol features, it can use these protocols across multiple paths that may have different performance and functional characteristics, and it can communicate with different Remote Endpoints to optimize performance, robustness to failure, or some other metric. Beyond these, if the API for the system remains the same over time, new protocols and features may be added to the system’s implementation without requiring changes in applications for adoption.
The following considerations were used in the design of this architecture.
Functionality that is common across multiple transport protocols should be accessible through a unified set of API calls. An application should be able to implement logic for its basic use of transport networking (establishing the transport, and sending and receiving data) once, and expect that implementation to continue to function as the transports change.
Any Transport Services API must allow access to the distilled minimal set of features offered by transport protocols [I-D.ietf-taps-minset].
There are applications that will need to control fine-grained details of transport protocols to optimize their behavior and ensure compatibility with remote peers,. A Transport Services system will therefore also needs to allow more specialized protocol features to be used. The interface for these specialized options should be exposed differently from the common options to ensure flexibility.
A specialized feature could be required by an application only when using a specific protocol, and not when using others. For example, if an application is using UDP, it could require control over the checksum or fragmentation behavior for UDP; if it used a protocol to frame its data over a byte stream like TCP, it would not need these options. In such cases, the API should expose the features in such a way that they take effect when a particular protocol is selected, but do not imply that only that protocol could be used if there are equivalent options.
Other specialized features, however, may be strictly required by an application and thus constrain the set of protocols that can be used. For example, if an application requires encryption of its transport data, only protocol stacks that include some transport security protocol are eligible to be used. A Transport Services API must allow applications to define such requirements and constrain the system’s options. Since such options are not part of the core/common features, it should be simple for an application to modify its set of constraints and change the set of allowable protocol features without changing the core implementation.
The Transport Services API is envisioned as the abstract model for a family of APIs that share a common way to expose transport features and encourage flexibility. The abstract API definition [I-D.ietf-taps-interface] describes this interface and is aimed at application developers.
Implementations that provide the Transport Services API [I-D.ietf-taps-impl] will vary due to system-specific support and the needs of the deployment scenario. It is expected that all implementations of Transport Services will offer the entire mandatory API, but that some features will not be functional in certain implementations. All implementations must offer sufficient APIs to use the distilled minimal set of features offered by transport protocols [I-D.ietf-taps-minset], including API support for TCP and UDP transport, but it is possible that some very constrained devices might not have, for example, a full TCP implementation.
To preserve flexibility and compatibility with future protocols, top-level features in the Transport Services API should avoid referencing particular transport protocols. The mappings of these API features to specific implementations of each feature is explained in the [TAPS-IMPL], which also explain the implications of the feature provided by existing protocols. It is expected that this document will be updated and supplemented as new protocols and protocol features are developed.
It is important to note that neither the Transport Services API nor the Implementation document defines new protocols that require any changes to a remote host. The Transport Services system must be deployable on one side only, as a way to allow an application to make better use of available capabilities on a system and protocol features that may be supported by peers across the network.
The concepts defined in this document are intended primarily for use in the documents and specifications that describe the Transport Services architecture and API. While the specific terminology may be used in some implementations, it is expected that there will remain a variety of terms used by running code.
The architecture divides the concepts for Transport Services into two categories:
The following diagram summarizes the top-level concepts in the architecture and how they relate to one another.
+-----------------------------------------------------+
| Application |
+-+----------------+------^-------+--------^----------+
| | | | |
pre- | data | events
establishment | transfer | |
| establishment | termination |
| | | | |
| +--v------v-------v+ |
+-v-------------+ Basic Objects +-------+----------+
| Transport +--------+---------+ |
| Services | |
| API | |
+------------------------|----------------------------+
|
+------------------------|----------------------------+
| Transport | |
| System | +-----------------+ |
| Implementation | | Cached | |
| | | State | |
| (Candidate Gathering) | +-----------------+ |
| | |
| (Candidate Racing) | +-----------------+ |
| | | System | |
| | | Policy | |
| +----------v-----+ +-----------------+ |
| | Protocol | |
+-------------+ Stack(s) +----------------------+
+-------+--------+
V
Network Layer Interface
Figure 1: Concepts and Relationships in the Transport Services Architecture
Fundamentally, a Transport Services API needs to provide basic objects (Section 4.1.1) that allow applications to establish communication and send and receive data. These may be exposed as handles or referenced objects, depending on the language.
Beyond the basic objects, there are several high-level groups of actions that any Transport Services API must provide:
The diagram below provides a high-level view of the actions taken during the lifetime of a connection.
Pre-Establishment : Established : Termination
----------------- : ----------- : -----------
: Close() :
+---------------+ Initiate() +------------+ Abort() :
+-->| Preconnection |----------->| Connection |---------------> Closed
| +---------------+ : +------------+ Connection:
| : ^ ^ | Finished :
+-- Local Endpoint : | | | :
| : | | +---------+ :
+-- Remote Endpoint : | | | :
| : | |Send() | :
+-- Path Selection : | +---------+ v :
| Properties : | | Message | Message :
| : | | to send | Received :
+-- Protocol Selection : | +---------+ :
| Properties : | :
| : | :
+-- Specific Protocol : | :
| Properties : | :
| : | :
| +----------+ : | :
+-->| Listener |-----------------+ :
+----------+ Connection Received :
^ : :
| : :
Listen() : :
Figure 2: The lifetime of a connection
This list of events that can be delivered to an application is not exhaustive, but gives the top-level categories of events. The API may expand this list.
The Transport System Implementation Concepts define the set of objects used internally to a system or library to provide the functionality required to provide a transport service across a network, as required by the abstract interface.
The Transport Services architecture defines a mechanism that allows applications to easily use different network paths and Protocol Stacks. Transitioning between different Protocol Stacks may in some cases be controlled by properties that only change when application code is updated. For example, an application may enable the use of a multipath or multistreaming transport protocol by modifying the properties in its Pre-Connection configuration. In some cases, however, the Transport Services system will be able to automatically change Protocol Stacks without an update to the application, either by selecting a new stack entirely, or racing multiple candidate Protocol Stacks during connection establishment. This functionality can be a powerful driver of new protocol adoption, but must be constrained carefully to avoid unexpected behavior that can lead to functional or security problems.
If two different Protocol Stacks can be safely swapped, or raced in parallel (see Section 4.2.2), then they are considered to be “equivalent”. Equivalent Protocol Stacks must meet the following criteria:
The inclusion of transport security protocols [I-D.ietf-taps-transport-security] in a Protocol Stack adds extra restrictions to Protocol Stack equivalence. Security features and properties, such as cryptographic algorithms, peer authentication, and identity privacy vary across security protocols, and across versions of security protocols. Protocol equivalence should not be assumed for different protocols or protocol versions, even if they offer similar application configuration options.
To ensure that security protocols are not incorrectly swapped, Transport Services systems should only automatically generate equivalent Protocol Stacks when the transport security protocols within the stacks are identical. Specifically, a system should consider protocols identical only if they are of the same type and version. For example, the same version of TLS running over two different transport protocol stacks may be considered equivalent, whereas TLS 1.2 and TLS 1.3 [I-D.ietf-tls-tls13] should not be considered equivalent.
While some transports expose a byte stream abstraction, most higher level protocols impose some structure onto that byte stream. That is, the higher level protocol operates in terms of messages, protocol data units (PDUs), rather than using unstructured sequences of bytes, with each message being processed in turn. Protocols are specified in terms of state machines acting on semantic messages, with parsing the byte stream into messages being a necessary annoyance, rather than a semantic concern. Accordingly, the Transport Services architecture exposes messages as the primary abstraction. Protocols that deal only in byte streams, such as TCP, represent their data in each direction as a single, long message. When framing protocols are placed on top of byte streams, the messages used in the API represent the framed messages within the stream.
Providing a message-based abstraction also provides:
All require explicit message boundaries, and application-level framing of messages, to be effective. Once a message is passed to the transport, it can not be cancelled or paused, but prioritization as well as lifetime and retransmission management will provide the protocol stack with all needed information to send the messages as quickly as possible without blocking transmission unnecessarily. The transport services architecture facilitates this by handling messages, with known identity (sequence numbers, in the simple case), lifetimes, niceness, and antecedents.
Transport protocols such as SCTP provide a message-oriented API that has similar features to those we describe. Other transports, such as TCP, do not. To support a message oriented API, while still being compatible with stream-based transport protocols, implementations of the transport services architecture should provide APIs for framing and de-framing messages. That is, we push message framing down into the transport services API, allowing applications to send and receive complete messages. This is backwards compatible with existing protocols and APIs, since the wire format of messages does not change, but gives the protocol stack additional information to allow it to make better use of modern transport services.
RFC-EDITOR: Please remove this section before publication.
This document has no actions for IANA.
The Transport Services architecture does not recommend use of specific security protocols or algorithms. Its goal is to offer ease of use for existing protocols by providing a generic security-related interface. Each provided interface mimics an existing protocol-specific interface provided by supported security protocols. For example, trust verification callbacks are common parts of TLS APIs. Transport Services APIs will expose similar functionality.
Clients must take care to use security APIs appropriately. In cases where clients use said interface to provide sensitive keying material, e.g., access to private keys or copies of pre-shared keys (PSKs), key use must be validated. For example, clients should not use PSK material created for the Encapsulating Security Protocol (ESP, part of IPsec) [RFC4303] with QUIC, and clients must not use private keys intended for server authentication as a keys for client authentication. Moreover, unlike certain transport features such as TCP Fast Open (TFO) [RFC7413] or Explicit Congestion Notification (ECN) [RFC3168] which can fall back to standard configurations, Transport Services systems must not permit fallback for security protocols. For example, if a client requests TLS, yet TLS or the desired version are not available, its connection must fail. Clients are responsible for implementing protocol or version fallback using a Transport Services API if so desired.
This work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 644334 (NEAT).
This work has been supported by Leibniz Prize project funds of DFG - German Research Foundation: Gottfried Wilhelm Leibniz-Preis 2011 (FKZ FE 570/4-1).
This work has been supported by the UK Engineering and Physical Sciences Research Council under grant EP/R04144X/1.
Thanks to Stuart Cheshire, Josh Graessley, David Schinazi, and Eric Kinnear for their implementation and design efforts, including Happy Eyeballs, that heavily influenced this work.