| Internet-Draft | TAPS Architecture | March 2020 |
| Pauly, et al. | Expires 10 September 2020 | [Page] |
This document describes an architecture for exposing transport protocol features to applications for network communication, the Transport Services architecture. The Transport Services Application Programming Interface (API) is based on an asynchronous, event-driven interaction pattern. It uses messages for representing data transfer to applications, and it assumes an implementation that can use multiple IP addresses, multiple protocols, and multiple paths, and provide multiple application streams. This document further defines common terminology and concepts to be used in definitions of Transport Services APIs and implementations.¶
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 10 September 2020.¶
Copyright (c) 2020 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 (Socket API). The naming of objects and functions across these APIs is not consistent, and varies depending on the protocol being used. For example, sending and receiving streams of data is conceptually the same for both an unencrypted Transmission Control Protocol (TCP) stream and operating on an encrypted Transport Layer Security (TLS) [RFC8446] 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 transport protocols varies based on the context of the protocols themselves: terms such as "flow", "stream", "message", and "connection" can take on many different meanings. 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, without requiring changes to the applications. This flexibility enables faster deployment of new features and protocols. It can also support applications by offering racing and fallback mechanisms, which otherwise need to be implemented in each application separately.¶
This document was developed in parallel with the specification of the Transport Services API [I-D.ietf-taps-interface] and Implementation Guidelines [I-D.ietf-taps-impl]. Although following the Transport Services Architecture does not require that all APIs and implementations are identical, a common minimal set of features represented in a consistent fashion will enable applications to be easily ported from one system to another.¶
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]. These documents 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 in that document as Functional or Optimizing Features), or else can be handled automatically by a system implementing Transport Services (referred to as Automatable Features). Among the Functional and Optimizing 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, but would not harm the functionality of other protocols. For example, some protocols can deliver messages faster for applications that do not require messages to arrive in the order in which they were sent. However, this functionality needs to be explicitly allowed by the application, since reordering messages would be undesirable in many cases.¶
This document describes the Transport Services architecture in three sections:¶
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.¶
The traditional model of using sockets for networking can be represented as follows:¶
+-----------------------------------------------------+
| Application |
+-----------------------------------------------------+
| | |
+------------+ +------------+ +--------------+
| stub | | Stream API | | Datagram API |
| resolver | +------------+ +--------------+
+------------+ | |
+---------------------------------+
| TCP UDP |
| Kernel Networking Stack |
+---------------------------------+
|
+-----------------------------------------------------+
| Network Layer Interface |
+-----------------------------------------------------+
The Transport Services architecture evolves this general model of interaction, aiming to both modernize the API surface presented to applications by the transport layer and enrich the capabilities of the Transport Services implementation. It combines interfaces for multiple interaction patterns into a unified whole. By combining name resolution with connection establishment and data transfer in a single API, it allows for more flexible implementations to provide path and transport protocol agility on the application's behalf.¶
+-----------------------------------------------------+
| Application |
+-----------------------------------------------------+
|
+-----------------------------------------------------+
| Transport Services API |
+-----------------------------------------------------+
|
+-----------------------------------------------------+
| Transport Services Implementation |
| (Using: DNS, 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 network connections and transfer data. The implementation [I-D.ietf-taps-impl] is responsible for mapping the API to the various available transport protocols and managing the available network interfaces and paths.¶
There are key differences between the architecture of the Transport Services system and the architecture of the Socket API: the Transport Services API is asynchronous and event-driven; it uses messages for representing 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. Emulation of an asynchronous interface using sockets generally uses 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 and can try again later.¶
In contrast to sockets, all interaction with a Transport Services system is expected to be asynchronous, and use an event-driven model (see Section 4.1.5). For example, if the application wants to read, its call to read will not complete immediately, but will deliver an event containing the received data once it is available. Error handling is also asynchronous; a failure to send results in an asynchronous send error as an event.¶
The Transport Services API also delivers events regarding the lifetime of a connection and changes in the available network links, which were not previously made explicit in sockets.¶
Using asynchronous events allows for a more natural 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.¶
Separate from events, callbacks are also provided for asynchronous interactions with the API not directly related to events on the network or network interfaces.¶
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. For example, HTTP/1.1 uses character delimiters to segment messages over a stream [RFC7230]; TLS record headers carry a version, content type, and length [RFC8446]; and HTTP/2 uses frames to segment its headers and bodies [RFC7540].¶
The Transport Services API represents data as messages, so that it more closely matches the way applications use the network. Providing a message-based abstraction provides many benefits, such as:¶
Allowing applications to interact with messages is backwards-compatible with existings protocols and APIs, as it does not change the wire format of any protocol. Instead, it gives the protocol stack additional information to allow it to make better use of modern transport services, while simplifying the application's role in parsing data. For protocols which natively use a streaming abstraction, framers (Section 4.1.4) bridge the gap between the two abstractions.¶
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. Transport services implementations can further implement fallback mechanisms if connection establishment of one protocol fails or performance is detected to be unsatisfactory.¶
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 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 is intended to 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 could have different performance and functional characteristics; and it can communicate with different remote systems 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 could be added to the system's implementation without requiring changes in applications for adoption.¶
Functionality that is common across multiple transport protocols ought to be accessible through a unified set of API calls. An application using a Transport Services API can 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.¶
As a baseline, any Transport Services API needs to 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 systems. A Transport Services system therefore ought to also permit more specialized protocol features to be used. The interface for these specialized options ought to 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 TCP, it could require control over the User Timeout Option for TCP; these options would not take effect for other transport protocols. In such cases, the API ought to 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. For example, if the API allows an application to specify a preference to use the User Timeout Option, communication would not fail when a protocol such as QUIC is selected.¶
Other specialized features, however, could be strictly required by an application and thus constrain the set of protocols that can be used. For example, if an application requires support for automatic handover or failover for a connection, only protocol stacks that provide this feature are eligible to be used, e.g., protocol stacks that include a multipath protocol or a protocol that supports connection migration. A Transport Services API needs to allow applications to define such requirements and constrain the system's options. Since such options are not part of the core/common features, it will generally 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 how it can be exposed to 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. All implementations are expected to offer an API that is sufficient 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. However, some features provided by this API will not be functional in certain implementations. For example, it is possible that some very constrained devices might not have a full TCP implementation beneath the API.¶
To preserve flexibility and compatibility with future protocols, top-level features in the Transport Services API ought to avoid referencing particular transport protocols. The mappings of these API features to specific implementations of each feature is explained in the [I-D.ietf-taps-impl] along with the implications of the feature on existing protocols. It is expected that [I-D.ietf-taps-interface] will be updated and supplemented as new protocols and protocol features are developed.¶
It is important to note that neither the Transport Services API [I-D.ietf-taps-interface] nor the Implementation document [I-D.ietf-taps-impl] define new protocols or protocol capabilities that affect what is communicated across the network. Use of a Transport Services system does not require that a peer on the other side of a connection uses the same API or implementation. A Transport Services system acting as a connection initiator can communicate with any existing system that implements the transport protocol(s) selected by the Transport Services system. Similarly, a Transport Services system acting as a listener can receive connections for any protocol that is supported by the system from existing initiators that implement the protocol, independent of whether the initiator uses Transport Services as well or not.¶
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 can 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-------------+ Connection(s) +-------+----------+
| Transport +--------+---------+ |
| Services | |
| API | +-------------+ |
+------------------------+--+ Framer(s) |-----------+
| +-------------+
+------------------------|----------------------------+
| Transport | |
| System | +-----------------+ |
| Implementation | | Cached | |
| | | State | |
| (Candidate Gathering) | +-----------------+ |
| | |
| (Candidate Racing) | +-----------------+ |
| | | System | |
| | | Policy | |
| +----------v-----+ +-----------------+ |
| | Protocol | |
+-------------+ Stack(s) +----------------------+
+-------+--------+
V
Network Layer Interface
Fundamentally, a Transport Services API needs to provide connection objects (Section 4.1.1) that allow applications to establish communication, and then send and receive data. These could be exposed as handles or referenced objects, depending on the language.¶
Beyond the connection objects, there are several high-level groups of actions that any Transport Services API needs to provide:¶
The diagram below provides a high-level view of the actions and events during the lifetime of a Connection object. Note that some actions are alternatives (e.g., whether to initiate a connection or to listen for incoming connections), while others are optional (e.g., setting Connection and Message Properties in Pre-Establishment) or have been omitted for brevity and simplicity.¶
Pre-Establishment : Established : Termination
----------------- : ----------- : -----------
: :
+-- Local Endpoint : Message :
+-- Remote Endpoint : Receive() | :
+-- Transport Properties : Send() | :
+-- Security Parameters : | :
| : | :
| InitiateWithSend() | Close() :
| +---------------+ Initiate() +-----+------+ Abort() :
+---+ Preconnection |------------->| Connection |-----------> Closed
+---------------+ Rendezvous() +------------+ :
Listen() | : | | :
| : | v :
v : | Connection :
+----------+ : | Ready :
| Listener |----------------------+ :
+----------+ Connection Received :
: :
Transport Properties: Transport Properties allow the application to express their requirements, prohibitions, and preferences and configure the Transport Services system. There are three kinds of Transport Properties:¶
The following categories of events can be delivered to an application:¶
A Connection Group is a set of Connections that share properties and caches. For multiplexing transport protocols, only Connections within the same Connection Group are allowed to be multiplexed together. An application can explicitly define Connection Groups to control caching boundaries, as discussed in Section 4.2.4.¶
This section defines the set of objects used internally to a system or library to implement the functionality needed to provide a transport service across a network, as required by the abstract interface.¶
Connection establishment attempts for a set of candidates may be performed simultaneously, synchronously, serially, or some combination of all of these. We refer to this process as racing, borrowing terminology from Happy Eyeballs [RFC8305].¶
The Transport Services architecture defines a mechanism that allows applications to easily make use of various network paths and Protocol Stacks without requiring major changes in application logic. In some cases, changing which Protocol Stacks or network paths are used will require updating the preferences expressed by the application that uses the Transport Services system. For example, an application can 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 by racing multiple candidate Protocol Stacks during connection establishment. This functionality in the API can be a powerful driver of new protocol adoption, but needs to 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 need to meet the following criteria:¶
By default, stored properties of the implementation, such as cached protocol state, cached path state, and heuristics, may be shared (e.g. across multiple connections in an application). This provides efficiency and convenience for the application, since the Transport Services implementation can automatically optimize behavior.¶
There are several reasons, however, that an application might want to explicitly isolate some Connections. These reasons include:¶
The Transport Services API can allow applications to explicitly define Connection Groups that force separation of Cached State and Protocol Stacks. For example, a web browser application might use Connection Groups with separate caches for different tabs in the browser to decrease linkability.¶
The interface to specify a Connection Group can expose fine-grained tuning for which properties and cached state is allowed to be shared with other Connections. For example, an application might want to allow sharing TCP Fast Open cookies across groups, but not TLS session state.¶
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 translates to 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 [I-D.ietf-taps-transport-security].¶
As described above in Section 4.2.3, if a Transport Services system races between two different Protocol Stacks, both SHOULD use the same security protocols and options. However, a Transport Services system MAY race different security protocols, e.g., if the application explicitly specifies that it considers them equivalent.¶
Applications need to ensure that they use security APIs appropriately. In cases where applications use an interface to provide sensitive keying material, e.g., access to private keys or copies of pre-shared keys (PSKs), key use needs to be validated. For example, applications ought not to use PSK material created for the Encapsulating Security Protocol (ESP, part of IPsec) [RFC4303] with QUIC, and applications ought not to use private keys intended for server authentication as keys for client authentication.¶
Moreover, Transport Services systems MUST NOT automatically fall back from secure protocols to insecure protocols, or to weaker versions of secure protocols. For example, if an application requests a specific version of TLS, but the desired version of TLS is not available, its connection will fail. Applications are thus responsible for implementing security protocol fallback or version fallback by creating multiple Transport Services Connections, if so desired. Alternatively, a Transport Services system MAY allow applications to specify that fallback to a specific other version of a protocol is allowed.¶
This work has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreements No. 644334 (NEAT) and No. 688421 (MAMI).¶
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 Theresa Enghardt, Max Franke, Mirja Kuehlewind, Jonathan Lennox, and Michael Welzl for the discussions and feedback that helped shape the architecture described here. Thanks as well to Stuart Cheshire, Josh Graessley, David Schinazi, and Eric Kinnear for their implementation and design efforts, including Happy Eyeballs, that heavily influenced this work.¶