| TSVWG | G. Fairhurst |
| Internet-Draft | University of Aberdeen |
| Intended status: Informational | C. Perkins |
| Expires: November 2, 2020 | University of Glasgow |
| May 01, 2020 |
Considerations around Transport Header Confidentiality, Network Operations, and the Evolution of Internet Transport Protocols
draft-ietf-tsvwg-transport-encrypt-15
To protect user data and privacy, Internet transport protocols have supported payload encryption and authentication for some time. Such encryption and authentication is now also starting to be applied to the transport protocol headers. This helps avoid transport protocol ossification by middleboxes, while also protecting metadata about the communication. Current operational practice in some networks inspect transport header information within the network, but this is no longer possible when those transport headers are encrypted. This document discusses the possible impact when network traffic uses a protocol with an encrypted transport header. It suggests issues to consider when designing new transport protocols or features. These considerations arise from concerns such as network operations, prevention of network ossification, enabling transport protocol evolution and respect for user privacy.
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 November 2, 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.
Transport protocols have supported end-to-end encryption of payload data for many years. Examples include Transport Layer Security (TLS) over TCP, Datagram TLS (DTLS) over UDP, Secure RTP, and TCPcrypt. Some of these also provide integrity protection of all, or part, of the transport header.
This end-to-end transport payload encryption brings many benefits in terms of providing confidentiality and protecting user privacy. The benefits have been widely discussed, for example in [RFC7624]. This document supports and encourages increased use of end-to-end payload encryption in transport protocols. The implications of protecting the transport payload data are therefore not further discussed in this document.
A further level of protection can be achieved by encrypting the entire network layer payload, including both the transport headers and the transport payload data. This does not expose any transport header information to devices in the network, and therefore also prevents modification along a network path. An example of encryption at the network layer is the IPsec Encapsulating Security Payload (ESP) [RFC4303] in tunnel mode. Virtual Private Networks (VPNs) typically also operate in this way. This form of encryption is not further discussed in this document.
There is also a middle ground, comprising transport protocols that encrypt some, or all, of the transport layer header information, in addition to encrypting the transport payload data. An example of such a protocol, that is now seeing widespread interest and deployment, is the QUIC transport protocol [I-D.ietf-quic-transport]. The encryption and authentication of transport header information can prevent unwanted modification of transport header information by network devices, reducing the risk of protocol ossification. It also reduces the amount of metadata about the progress of the transport connection that is visible to the network [RFC8558]. In this document, the term "transport header information" is used to describe transport layer information concerning the operation of the transport protocol (i.e., information used by the transport protocol that might be carried in a protocol header). This does not refer to transport payload data (i.e., information transferred by the transport service), which itself could be encrypted.
The direction in which the use of transport header encryption evolves could have significant implications on the way the Internet architecture develops, and therefore needs to be considered as a part of protocol design and evolution. This includes considering whether the endpoints permit (or are able to permit) network devices to observe specific information by explicitly exposing a transport header field (or a field derived from transport header information) to the network; whether it is intended that a network device can modify a transport header field; and whether any modification along the network path can be detected by the receiving endpoint. This can require changes to network operations and other practises and could drive changes to the design of network measurement for research, operational, and standardisation purposes.
As discussed in [RFC7258], the IETF has concluded that Pervasive Monitoring (PM) is a technical attack that needs to be mitigated in the design of IETF protocols, but RFC7528 also notes that "Making networks unmanageable to mitigate PM is not an acceptable outcome, but ignoring PM would go against the consensus documented here. An appropriate balance will emerge over time as real instances of this tension are considered". In support of achieving that balance, this document discusses design and deployment considerations for use of transport header encryption to protect against pervasive monitoring.
The transport protocols developed for the Internet are used across a wide range of paths across network segments with many different regulatory, commercial, and engineering considerations. This document considers some of the costs and changes to network management and research that are implied by widespread use of transport protocols that encrypt their transport header information. It reviews the implications of developing transport protocols that use end-to-end encryption to provide confidentiality of their transport layer headers, and considers the effect of such changes on transport protocol design, transport protocol evolution, and network operations. It also considers some anticipated implications on application evolution. This provides considerations relating to the design of transport protocols and features where the transport protocol encrypts some or all of their header information.
The transport layer provides end-to-end interactions between endpoints (processes) using an Internet path. Transport protocols layer over the network-layer service, and are usually sent in the payload of network-layer packets. Transport protocols support end-to-end communication between applications, using higher-layer protocols running on the end systems (i.e., transport endpoints).
This simple architectural view does not present one of the core functions of an Internet transport: to discover and adapt to the characteristics of the network path that is currently being used. The design of Internet transport protocols is as much about trying to avoid the unwanted side effects of congestion on a flow and other capacity-sharing flows, avoiding congestion collapse, adapting to changes in the path characteristics, etc., as it is about end-to-end feature negotiation, flow control, and optimising for performance of a specific application.
Transport headers have end-to-end meaning, but have often been observed by equipment within the network. Transport protocol specifications have not tended to consider this, and have often failed to indicate what parts of the transport header are intended to be invariant across protocol versions and visible to the network; to specify what parts of the transport header can be modified by the network to signal to the transport, and in what way; and to define which parts of the header are private and/or expected to change in future and which need to be protected for privacy or to prevent protocol ossification. This motivates a need to change the way transport protocols are designed, modified, and specified.
Increasing concern about pervasive network monitoring [RFC7258][RFC7624], and growing awareness of the problem of protocol ossification caused by middlebox interference with Internet traffic, has motivated a shift in transport protocol design. Recent transport protocols, such as QUIC [I-D.ietf-quic-transport], encrypt the majority of their transport headers to prevent observation and protect against modification by the network, and to make explicit their invariants and what is intended to be visible to the network.
Transport header encryption is expected to form a core part of future transport protocol designs. It can help to protect against pervasive monitoring, improve privacy, and reduce protocol ossification. Transport protocols that use header encryption with secure key distribution can provide confidentiality and protection for some, or all, of the transport header, controlling what is visible to, and can be modified by the network.
The increased use of transport header encryption has benefits, but also has implications for the broader ecosystem. The transport community has, to date, relied on measurements and insights from the network operations community to understand protocol behaviour, and to inform the selection of appropriate mechanisms to ensure a safe, reliable, and robust Internet. In turn, network operators and access providers have relied upon being able to observe traffic patterns and requirements, both in aggregate and at the flow level, to help understand and optimise the behaviour of their networks.
Transport header encryption can be used to intentionally limit the information available to network observers. The widespread use would therefore limit such observations, unless transport protocols are modified to selectively expose transport header information outside of the encrypted transport header. It is important to understand how transport header information is used by networks, to allow future protocol designs to make an informed choice on what, if any, transport layer information to expose to the network.
In-network measurement of transport flow characteristics can be used to enhance performance, control cost and improve service reliability. To support network operations and enhance performance, some operators have deployed functionality that utilises on-path observations of the transport headers of packets passing through their network ([RFC8517] gives an operator perspective on such use).
When network devices rely on the presence of a header field or the semantics of specific header information, this can lead to ossification where an endpoint has to supply a specific header to receive the network service that it desires.
In some cases, network-layer use of transport layer information can be benign or advantageous to the protocol (e.g., recognising the start of a TCP connection, providing header compression for a Secure RTP (SRTP) flow [RFC3711], or explicitly using exposed protocol information to provide consistent decisions by on-path devices). Header compression (e.g., [RFC5795]) depends on understanding of transport header and the way fields change packet-by-packet; as also do techniques to improve TCP performance by transparent modification of acknowledgement traffic [RFC3449]. Introducing a new transport protocol or changes to existing transport header information prevent these methods being used or require the network devices to be updated.
However, in other cases, ossification can have unwanted outcomes. Ossification can frustrate the evolution of a transport protocol. A mechanism implemented in a network device, such as a firewall, that requires a header field to have only a specific known set of values can prevent the device from forwarding packets using a different version of the protocol that introduces a feature that changes to a new value for the observed field.
An example of this type ossification was observed in the development of Transport Layer Security (TLS) 1.3 [RFC8446], where the design needed to function in the presence of deployed middleboxes that relied on the presence of certain header fields exposed in TLS 1.2. The design of MPTCP also had to be revised to account for middleboxes (known as "TCP Normalizers") that monitor the evolution of the window advertised in the TCP header and then reset connections when the window did not grow as expected. Similarly, issues have been reported using TCP. For example, TCP Fast Open [RFC7413] can experience middleboxes that modify the transport header of packets by removing "unknown" TCP options, segments with unrecognised TCP options can be dropped, segments that contain data and set the SYN bit can be dropped, or middleboxes that disrupt connections that send data before completion of the three-way handshake.
Other examples of ossification have included middleboxes that modify transport headers by rewriting TCP sequence and acknowledgement numbers, but are unaware of the (newer) TCP selective acknowledgement (SACK) option and therefore fail to correctly rewrite the SACK information to match the changes that were made to the fixed TCP header, preventing SACK from operating correctly.
In all these cases, middleboxes with a hard-coded, but incomplete, understanding of transport behaviour, interacted poorly with transport protocols after the transport behaviour was changed. In some case, the middleboxes modified or replaced information in the transport protocol header.
Transport header encryption prevents an on-path device from observing the transport headers, and therefore stops mechanisms being built that directly rely on or infer semantics of the transport header information. Encryption is normally combined with authentication of the protected information. RFC 8546 summarises this approach, stating that it is "The wire image, not the protocol's specification, determines how third parties on the network paths among protocol participants will interact with that protocol" [RFC8546], and it can be expected that header information that is not encrypted will become ossified.
While encryption can reduce ossification of the transport protocol, it does not itself prevent ossification of the network service. People seeking to understand network traffic could still come to rely on pattern inferences and other heuristics or machine learning to derive measurement data and as the basis for network forwarding decisions [RFC8546]. This can also create dependencies on the transport protocol, or the patterns of traffic it can generate, also in time resulting in ossification of the service.
The designers of a transport protocol decide whether to encrypt all, or a part of, the transport layer information. Section 4 of RFC8558 states: "Anything exposed to the path should be done with the intent that it be used by the network elements on the path" [RFC8558]. Protocol designs can decide not to encrypt certain transport header fields, making those fields observable in the network, or can define new fields designed to explicitly expose observable transport layer information to the network. Where exposed fields are intended to be immutable (i.e., can be observed, but not modified by a network device), the endpoints are encouraged to use authentication to provide a cryptographic integrity check that can detect if these immutable fields have been modified by network devices. Authentication can also help to prevent attacks that rely on sending packets that fake exposed control signals in transport headers (e.g., TCP RST spoofing).
Making part of a transport header observable, or exposing new header fields, can lead to ossification of that part of a header as network devices come to rely on observations of the exposed fields. A protocol design that provides an observable field might want to restrict the choice of usable values in a field by intentionally varying the format and/or value of the field to reduce the chance of ossification (see Section 4).
Transport header fields have been observed within the network for a variety of purposes. Some purposes are related to network management and operations. Use cases where the network devices intentionally modify mutable transport layer information are out of scope and are not described further in this document. More information may be found in other RFCs (e.g., [RFC3449], [RFC3135], [RFC8404], [RFC8462]), and [RFC8517].
The list below provides an overview with different uses of exposed immutable information.
This analysis does not judge whether specific practises are necessary. It is not an endorsement of any particular practice.
In response to pervasive monitoring [RFC7624] revelations and the IETF consensus that "Pervasive Monitoring is an Attack" [RFC7258], efforts are underway to increase encryption of Internet traffic. Applying confidentiality to transport header fields affects how protocol information is used [RFC8404], requiring consideration of the trade-offs discussed in Section 2.3.
The decision about which transport headers fields are made observable offers trade-offs around header confidentiality versus header observability (including non-encrypted, but authenticated, header fields) for network operations and management, and the implications for ossification and user privacy [Measurement]. Different parties will view the relative importance of these differently. For some, the benefits of encrypting all transport headers outweigh the impact of doing so; others might analyse the security, privacy and ossification impacts, and arrive at a different trade-off.
This section reviews examples of the observation of transport layer headers within the network. It does not consider intentional modification of transport headers by middleboxes (such as in Network Address Translation, NAT, or Firewalls). Common issues concerning IP address sharing are described in [RFC6269].
In-network observation of transport protocol headers requires knowledge of the format of the transport header:
The following subsections describe various ways that observable transport header information has been utilised.
Flow/Session identification [RFC8558] is a common function performed, for example, by measurement activities, QoS classifiers, firewalls, and as part of Denial of Service (DoS) prevention.
Observable transport header information, together with information in the network header, has been used to identify flows and their connection state, together with the set of protocol options being used. Transport protocols, such as TCP and the Stream Control Transport Protocol (SCTP), specify a standard base header that includes sequence number information and other data. They also have the possibility to negotiate additional headers at connection setup, identified by an option number in the transport header.
In some uses, an assigned transport port (e.g., 0..49151) can identify the protocol [RFC7605]. However, port information alone is not sufficient to guarantee identification. Applications can use arbitrary ports and do not need to use assigned port numbers. The use of an assigned port number is also not limited to the protocol for which the port is intended. Multiple sessions can also be multiplexed on a single port, and ports can be re-used by subsequent sessions.
Some flows can be identified by observing signalling protocol data (e.g., [RFC3261], [I-D.ietf-rtcweb-overview]) or through the use of magic numbers placed in the first byte(s) of the datagram payload [RFC7983].
When transport header information can not be observed, this removes information that could have been used to classify flows by passive observers along the path. More ambitious ways could be used to collect, estimate, or infer flow information, including heuristics based on the analysis of traffic patterns. For example, an operator that cannot access the Session Description Protocol (SDP) session descriptions to classify a flow as audio traffic, might instead use (possibly less-reliable) heuristics to infer that short UDP packets with regular spacing carry audio traffic. Operational practises aimed at inferring transport parameters are out of scope for this document, and are only mentioned here to recognise that encryption does not prevent operators from attempting to apply practises that were used with unencrypted transport headers. The IAB have provided a summary of expected implications of increased encryption on network functions that use the observable headers [RFC8546] and describe the expected benefits of designs that explicitly declare protocol invariant header information that can be used for this purpose.
Observable transport headers enable explicit measurement and analysis of protocol performance, network anomalies, and failure pathologies at any point along the Internet path. Some operators use passive monitoring to manage their portion of the Internet by characterising the performance of link/network segments. Inferences from transport headers are used to derive performance metrics. A variety of open source and commercial tools have been deployed that utilise transport header information in this way to derive the following metrics:
These metrics can support network operations, inform capacity planning, and assist in determining the demand for equipment and/or configuration changes by network operators. They can also inform Internet engineering activities by informing the development of new protocols, methodologies, and procedures.
In some cases, measurements could involve active injection of test traffic to perform a measurement (see Section 3.4 of [RFC7799]). However, most operators do not have access to user equipment, therefore the point of test is normally different from the transport endpoint. Injection of test traffic can incur an additional cost in running such tests (e.g., the implications of capacity tests in a mobile network are obvious). Some active measurements [RFC7799] (e.g., response under load or particular workloads) perturb other traffic, and could require dedicated access to the network segment.
Passive measurements (see Section 3.6 of [RFC7799]) can have advantages in terms of eliminating unproductive test traffic, reducing the influence of test traffic on the overall traffic mix, and the ability to choose the point of observation (see Section 3.2.1). Measurements can rely on observing packet headers, which is not possible if those headers are encrypted, but could utilise information about traffic volumes or patterns of interaction to deduce metrics.
One alternative approach is to use in-network techniques, which does not require the cooperation of an endpoint (see Section 5).
Information from the transport header is used by a multi-field classifier as a part of policy framework. Policies are commonly used for management of the QoS or Quality of Experience (QoE) in resource-constrained networks, and by firewalls to implement access rules (see also Section 2.2.2 of [RFC8404]). Network-layer classification methods that rely on a multi-field classifier (e.g., inferring QoS from the 5-tuple or choice of application protocol) are incompatible with transport protocols that encrypt the transport header information. Traffic that cannot be classified typically receives a default treatment.
Transport layer information can also be explicitly carried in network-layer header fields that are not encrypted, serving as a replacement/addition to the exposed transport header information [RFC8558]. This information can enable a different forwarding treatment by the network, even when a transport employs encryption to protect other header information.
The user of a transport that multiplexes multiple sub-flows might want to obscure the presence and characteristics of these sub-flows. On the other hand, an encrypted transport could set the network-layer information to indicate the presence of sub-flows, and to reflect the service requirements of individual sub-flows. There are several ways this could be done:
When transport headers cannot be observed, operators are unable to use this information directly. Careful use of the network layer features can help provide similar information in the case where the network is unable to inspect transport protocol headers. Section 6 describes use of network extension headers.
The common language between network operators and application/content providers/users is packet transfer performance at a layer that all can view and analyse. For most packets, this has been the transport layer, until the emergence of transport protocols performing header encryption, with the obvious exception of VPNs and IPsec.
When encryption hides more layers in each packet, people seeking understanding of the network operation rely more on pattern inference and other heuristics. It remains to be seen whether more complex inferences can be mastered to produce the same monitoring accuracy (see Section 2.1.1 of [RFC8404]).
When measurement datasets are made available by servers or client endpoints, additional metadata, such as the state of the network, is often necessary to interpret this data to answer questions about network performance or understand a pathology. Collecting and coordinating such metadata is more difficult when the observation point is at a different location to the bottleneck/device under evaluation [RFC7799].
Packet sampling techniques are used to scale the processing involved in observing packets on high rate links. This exports only the packet header information of (randomly) selected packets. The utility of these measurements depends on the type of bearer and number of mechanisms used by network devices. Simple routers are relatively easy to manage, but a device with more complexity demands understanding of the choice of many system parameters. This level of complexity exists when several network methods are combined.
This section discusses topics concerning observation of transport flows, with a focus on transport measurement.
On-path measurements are particularly useful for locating the source of problems, or to assess the performance of a network segment or a particular device configuration. Often issues can only be understood in the context of the other flows that share a particular path, common network device, interface port, etc. A simple example is monitoring of a network device that uses a scheduler or active queue management technique [RFC7567], where it could be desirable to understand whether the algorithms are correctly controlling latency, or if overload protection is working. This understanding implies knowledge of how traffic is assigned to any sub-queues used for flow scheduling, but can also require information about how the traffic dynamics impact active queue management, starvation prevention mechanisms, and circuit-breakers.
Sometimes multiple on-path observation points have to be used. By correlating observations of headers at multiple points along the path (e.g., at the ingress and egress of a network segment), an observer can determine the contribution of a portion of the path to an observed metric, to locate a source of delay, jitter, loss, reordering, congestion marking, etc.
Traffic rate and volume measurements are used by operators to help plan deployment of new equipment and configuration in their networks. Data is also valuable to equipment vendors who want to understand traffic trends and patterns of usage as inputs to decisions about planning products and provisioning for new deployments. This measurement information can also be correlated with billing information when this is also collected by an operator.
Trends in aggregate traffic can be observed and can be related to the endpoint addresses being used, but when transport header information is not observable, it might be impossible to correlate patterns in measurements with changes in transport protocols. This increases the dependency on other indirect sources of information to inform planning and provisioning.
Performance measurements (e.g., throughput, loss, latency) can be used by various actors to analyse the service offered to the users of a network segment, and to inform operational practice.
The traffic that can be observed by on-path network devices (the "wire image") is a function of transport protocol design/options, network use, applications, and user characteristics. In general, when only a small proportion of the traffic has a specific (different) characteristic, such traffic seldom leads to operational concern, although the ability to measure and monitor it is less. The desire to understand the traffic and protocol interactions typically grows as the proportion of traffic increases in volume. The challenges increase when multiple instances of an evolving protocol contribute to the traffic that share network capacity.
Operators can manage traffic load (e.g., when the network is severely overloaded) by deploying rate-limiters, traffic shaping, or network transport circuit breakers [RFC8084]. The information provided by observing transport headers is a source of data that can help to inform such mechanisms.
Transport header information can be utilised for a variety of operational tasks [RFC8404]: to diagnose network problems, assess network provider performance, evaluate equipment or protocol performance, capacity planning, management of security threats (including DoS), and responding to user performance questions. Section 3.1.2 and Section 5 of [RFC8404] provide further examples.
Operators can monitor the health of a portion of the Internet, to provide early warning and trigger action. Traffic and performance measurements can assist in setting buffer sizes, debugging and diagnosing the root causes of faults that concern a particular user's traffic. They can also be used to support post-mortem investigation after an anomaly to determine the root cause of a problem. In other cases, measurement involves dissecting network traffic flows. Observed transport header information can help identify whether link/network tuning is effective and alert to potential problems that can be hard to derive from link or device measurements alone.
An alternative could rely on access to endpoint diagnostic tools or user involvement in diagnosing and troubleshooting unusual use cases or to troubleshoot non-trivial problems.
Another approach is to use traffic pattern analysis. Such tools can provide useful information during network anomalies (e.g., detecting significant reordering, high or intermittent loss), however indirect measurements would need to be carefully designed to provide information for diagnostics and troubleshooting.
The design trade-offs for radio networks are often very different from those of wired networks. A radio-based network (e.g., cellular mobile, enterprise WiFi, satellite access/back-haul, point-to-point radio) has the complexity of a subsystem that performs radio resource management, with direct impact on the available capacity, and potentially loss/reordering of packets. The impact of the pattern of loss and congestion, differs for different traffic types, correlation with propagation and interference can all have significant impact on the cost and performance of a provided service. For radio links, the use for this type of information is expected to increase as operators bring together heterogeneous types of network equipment and seek to deploy opportunistic methods to access radio spectrum.
Lack of tools and resulting information can reduce the ability of an operator to observe transport performance and could limit the ability of network operators to trace problems, make appropriate QoS decisions, or respond to other queries about the network service.
A network operator supporting traffic that uses transport header encryption is unable to use tools that rely on transport protocol information. However, the use of encryption has the desirable effect of preventing unintended observation of the payload data and these tools seldom seek to observe the payload, or other application details. A flow that hides its transport header information could imply "don't touch" to some operators. This might limit a trouble-shooting response to "can't help, no trouble found".
Header compression saves link capacity by compressing network and transport protocol headers on a per-hop basis. It was widely used with low bandwidth dial-up access links, and still finds application on wireless links that are subject to capacity constraints. Examples of header compression include use with TCP/IP and RTP/UDP/IP flows [RFC2507], [RFC2508], [RFC4995].
While it is possible to compress only the network layer headers, significant savings can be made if both the network and transport layer headers are compressed together as a single unit. The SRTP extensions [RFC3711] were explicitly designed to leave the transport protocol headers unencrypted, but authenticated, since support for header compression was considered important. Encrypting the transport protocol headers does not break such header compression, but does cause a fall back to compressing only the network layer headers, with a significant reduction in efficiency.
End-to-end encryption can be applied at various protocol layers. It can be applied above the transport to encrypt the transport payload (e.g., using TLS). This can hide information from an eavesdropper in the network. It can also help protect the privacy of a user, by hiding data relating to user/device identity or location.
There are several motivations for encryption:
The use of transport header authentication and encryption exposes a tussle between middlebox vendors, operators, applications developers and users:
A decision to use transport header encryption can improve user privacy, and can reduce protocol ossification and help the evolution of the transport protocol stack, but is also has implications for network operations and management.
The following briefly reviews some security design options for transport protocols. A Survey of Transport Security Protocols [I-D.ietf-taps-transport-security] provides more details concerning commonly used encryption methods at the transport layer.
As seen, different transports use encryption to protect their header information to varying degrees. The trend is towards increased protection.
An on-path device can make measurements by utilising additional protocol headers carrying operations, administration and management (OAM) information in an additional packet header. Using network-layer approaches to reveal information has the potential that the same method (and hence same observation and analysis tools) can be consistently used by multiple transport protocols. This approach also could be applied to methods beyond OAM (see Section 6). There can also be less desirable implications from separating the operation of the transport protocol from the measurement framework.
OAM information can be added at the ingress to a maintenance domain (e.g., an Ethernet protocol header with timestamps and sequence number information using a method such as 802.11ag or in-situ OAM [I-D.ietf-ippm-ioam-data], or as a part of encapsulation protocol). The additional header information is typically removed the at the egress of the maintenance domain.
Although some types of measurements are supported, this approach does not cover the entire range of measurements described in this document. In some cases, it can be difficult to position measurement tools at the appropriate segments/nodes and there can be challenges in correlating the downstream/upstream information when in-band OAM data is inserted by an on-path device.
OAM information can also be added at the network layer as an IPv6 extension header or an IPv4 option. This information can be used across multiple network segments, or between the transport endpoints.
One example is the IPv6 Performance and Diagnostic Metrics (PDM) destination option [RFC8250]. This allows a sender to optionally include a destination option that caries header fields that can be used to observe timestamps and packet sequence numbers. This information could be authenticated by receiving transport endpoints when the information is added at the sender and visible at the receiving endpoint, although methods to do this have not currently been proposed. This method has to be explicitly enabled at the sender.
A transport protocol can choose to expose certain transport information to on-path devices operating at the network layer by sending observable fields. One approach is to make an explicit choice not to encrypt certain transport header fields, making this transport information observable by the network. Another approach is to choose to expose transport information through the use of network-layer extension headers. Both are examples of explicit information intended to be used by network devices on the path [RFC8558].
Whatever the mechanism used to expose the information, a decision to only expose specific transport information, places the transport endpoint in control of what to expose or not to expose outside of the encrypted transport header. This decision can then be made independently of the transport protocol functionality. This can be done by exposing part of the transport header or as a network layer option/extension.
At the network-layer, packets can carry optional headers (similar to Section 5) that may be used to explicitly expose transport header information to the on-path devices operating at the network layer (Section 3.1.3). For example, an endpoint that sends an IPv6 Hop-by-Hop option [RFC8200] can provide explicit transport layer information that can be observed and used by network devices on the path.
Network-layer optional headers explicitly indicate the information that is exposed, whereas use of exposed transport header information first requires an observer to identify the transport protocol and its format. See Section 3.1.1 for further discussion of transport protocol identification.
An arbitrary path can include one or more network devices that drop packets that include a specific header or option used for this purpose (see [RFC7872]). This could impact the proper functioning of the protocols using the path. Protocol methods can be designed to probe to discover whether the specific option(s) can be used along the current path, enabling use on arbitrary paths.
There are opportunities for multiple transport protocols to consistently supply common observable information [RFC8558]. A common approach can result in an open definition of the observable fields. This has the potential that the same information can be utilised across a range of operational and analysis tools.
The motivation to reflect actual transport header information and the implications of network devices using this information has to be considered when proposing such a method. RFC8558 summarises this as "When signals from endpoints to the path are independent from the signals used by endpoints to manage the flow's state mechanics, they may be falsified by an endpoint without affecting the peer's understanding of the flow's state. For encrypted flows, this divergence is not detectable by on-path devices." [RFC8558].
Considerations concerning the selection of appropriate information to expose include:
The choice of which transport header fields to expose and which to encrypt is a design decision for the transport protocol. Selective encryption requires trading conflicting goals of observability and network support, privacy, and risk of ossification, to decide what header fields to protect and which to make visible.
Security work typically employs a design technique that seeks to expose only what is needed. This approach provides incentives to not reveal any information that is not necessary for the end-to-end communication. However, there can be performance and operational benefits in exposing selected information to network tools.
This section explores key implications of working with encrypted transport protocols.
Independent observation by multiple actors is important if the transport community is to maintain an accurate understanding of the network. Encrypting transport header encryption changes the ability to collect and independently analyse data. Internet transport protocols employ a set of mechanisms. Some of these have to work in cooperation with the network layer for loss detection and recovery, congestion detection and control. Others have to work only end-to-end (e.g., parameter negotiation, flow-control).
The majority of present Internet applications use two well-known transport protocols, TCP and UDP. Although TCP represents the majority of current traffic, many real-time applications use UDP, and much of this traffic uses RTP format headers in the payload of the UDP datagram. Since these protocol headers have been fixed for decades, a range of tools and analysis methods have became common and well-understood.
Protocols that expose the state information used by the transport protocol in their header information (e.g., timestamps used to calculate the RTT, packet numbers used to asses congestion and requests for retransmission) provide an incentive for the sending endpoint to provide correct information, since the protocol will not work otherwise. This increases confidence that the observer understands the transport interaction with the network. For example, when TCP is used over an unencrypted network path (i.e., one that does not use IPsec or other encryption below the transport), it implicitly exposes transport header information that can be used for measurement at any point along the path. This information is necessary for the protocol's correct operation, therefore there is no incentive for a TCP or RTP implementation to put incorrect information in this transport header. A network device can have confidence that the well-known (and ossified) transport header information represents the actual state of the endpoints.
When encryption is used to hide some or all of the transport headers, the transport protocol chooses which information to reveal to the network about its internal state, what information to leave encrypted, and what fields to grease to protect against future ossification. Such a transport could be designed (or an existing transport modified), for example, to provide summary data regarding its performance, congestion control state, etc., or to make available explicit measurement information. For example, a QUIC endpoint can optionally set the spin bit to reflect to explicitly reveal the RTT of an encrypted transport session to the on-path network devices [I-D.ietf-quic-transport]).
When providing or using such information, it is important to consider the privacy of the user and their incentive for providing accurate and detailed information. Protocols that selectively reveal some transport state or measurable information are choosing to establish a trust relationship with the network operators. There is no protocol mechanism that can guarantee that the information provided represents the actual transport state of the endpoints, since those endpoints can always send additional information in the encrypted part of the header, to update or replace whatever they reveal. This reduces the ability to independently measure and verify that a protocol is behaving as expected. For some operational uses, the information has to contain sufficient detail to understand, and possibly reconstruct, the network traffic pattern for further testing. In this case, operators have to gain the trust of transport protocol implementers if the transport headers are to correctly reveal such information.
OAM data records [I-D.ietf-ippm-ioam-data] could be embedded into a variety of encapsulation methods at different layers to support the goals of a specific operational domain. OAM-related metadata can support functions such as performance evaluation, path-tracing, path verification information, classification and a diversity of other uses. When encryption is used to hide some or all of the transport headers, analysis requires coordination between actors at different layers to successfully characterise flows and correlate the performance or behaviour of a specific mechanism with the configuration and traffic using operational equipment (e.g., combining transport and network measurements to explore congestion control dynamics, the implications of designs for active queue management or circuit breakers).
Some measurements could be completed by utilising endpoint-based logging (e.g., based on Quic-Trace). Such information has a diversity of uses, including developers wishing to debug/understand the transport/application protocols with which they work, researchers seeking to spot trends and anomalies, and to characterise variants of protocols. A standard format for endpoint logging could allow these to be shared (after appropriate anonymisation) to understand performance and pathologies. Measurements based on logging have to establish the validity and provenance of the logged information to establish how and when traces were captured.
Despite being applicable in some scenarios, endpoint logs do not provide equivalent information to in-network measurements. In particular, endpoint logs contain only a part of the information to understand the operation of network devices and identify issues such as link performance or capacity sharing between multiple flows. Additional information has to be combined to determine which equipment/links are used and the configuration of equipment along the network paths being measured.
The patterns and types of traffic that share Internet capacity change over time as networked applications, usage patterns and protocols continue to evolve.
If "unknown" or "uncharacterised" traffic patterns form a small part of the traffic aggregate passing through a network device or segment of the network the path, the dynamics of the uncharacterised traffic might not have a significant collateral impact on the performance of other traffic that shares this network segment. Once the proportion of this traffic increases, monitoring the traffic can determine if appropriate safety measures have to be put in place.
Tracking the impact of new mechanisms and protocols requires traffic volume to be measured and new transport behaviours to be identified. This is especially true of protocols operating over a UDP substrate. The level and style of encryption has to be considered in determining how this activity is performed. On a shorter timescale, information could also have to be collected to manage DoS attacks against the infrastructure.
Information provided by tools observing transport headers can be used to classify traffic, and to limit the network capacity used by certain flows, as discussed in Section 3.2.4). Equally, operators could use analysis of transport headers and transport flow state to demonstrate that they are not providing differential treatment to certain flows. Obfuscating or hiding this information using encryption could lead operators and maintainers of middleboxes (firewalls, etc.) to seek other methods to classify, and potentially other mechanisms to condition, network traffic.
A lack of data that reduces the level of precision with which flows can be classified also reduces the design space for conditioning mechanisms (e.g., rate limiting, circuit breaker techniques [RFC8084], or blocking of uncharacterised traffic), and this has to be considered when evaluating the impact of designs for transport encryption [RFC5218].
Some network operators currently use observed transport header information as a part of their operational practice, and have developed tools and techniques that use information observed in currently deployed transports and their applications. A variety of open source and proprietary tools have been deployed that use this information for a variety of short and long term measurements. Encryption of the transport header information prevents tooling from directly observing the transport header information, limiting its utility.
Alternative diagnostic and troubleshooting tools would have to be developed and deployed is transport header encryption is widely deployed. Introducing a new protocol or application might then require these tool chains and practises to be updated, and could in turn impact operational mechanisms, and policies. Each change can introduce associated costs, including the cost of collecting data, and the tooling to handle multiple formats (possibly as these co-exist in the network, when measurements span time periods during which changes are deployed, or to compare with historical data). These costs are incurred by an operator to manage the service and debug network issues.
At the time of writing, the overall operational impact of using encrypted transports is not yet well understood. Design trade-offs could mitigate these costs by explicitly choosing to expose selected information (e.g., header invariants and the spin-bit in QUIC [I-D.ietf-quic-transport]), the specification of common log formats, and development of alternative approaches.
Transport protocol evolution, and the ability to measure and understand the impact of protocol changes, have to proceed hand-in-hand. A transport protocol that provides observable headers can be used to provide open and verifiable measurement data. Observation of pathologies has a critical role in the design of transport protocol mechanisms and development of new mechanisms and protocols. This helps understanding the interactions between cooperating protocols and network mechanism, the implications of sharing capacity with other traffic and the impact of different patterns of usage. The ability of other stake holders to review transport header traces helps develop insight into performance and traffic contribution of specific variants of a protocol.
Development of new transport protocol mechanisms has to consider the scale of deployment and the range of environments in which the transport is used. Experience has shown that it is often difficult to correctly implement new mechanisms [RFC8085], and that mechanisms often evolve as a protocol matures, or in response to changes in network conditions, changes in network traffic, or changes to application usage. Analysis is especially valuable when based on the behaviour experienced across a range of topologies, vendor equipment, and traffic patterns.
New transport protocol formats are expected to facilitate an increased pace of transport evolution, and with it the possibility to experiment with and deploy a wide range of protocol mechanisms. There has been recent interest in a wide range of new transport methods, e.g., Larger Initial Window, Proportional Rate Reduction (PRR), congestion control methods based on measuring bottleneck bandwidth and round-trip propagation time, the introduction of AQM techniques and new forms of ECN response (e.g., Data Centre TCP, DCTP, and methods proposed for L4S). The growth and diversity of applications and protocols using the Internet also continues to expand. For each new method or application it is desirable to build a body of data reflecting its behaviour under a wide range of deployment scenarios, traffic load, and interactions with other deployed/candidate methods.
Encryption of transport header information could reduce the range of actors that can observe useful data. This would limit the information sources available to the Internet community to understand the operation of new transport protocols, reducing information to inform design decisions and standardisation of the new protocols and related operational practises. The cooperating dependence of network, application, and host to provide communication performance on the Internet is uncertain when only endpoints (i.e., at user devices and within service platforms) can observe performance, and when performance cannot be independently verified by all parties.
Independently observed data is also important to ensure the health of the research and development communities and can help promote acceptance of proposed specifications by the wider community (e.g., as a method to judge the safety for Internet deployment) and provides valuable input during standardisation. Open standards motivate a desire to include independent observation and evaluation of performance data, which in turn demands control/understanding about where and when measurement samples are collected. This requires consideration of the methods used to observe data and the appropriate balance between encrypting all and no transport header information.
Header encryption and strong integrity checks are being incorporated into new transport protocols and have important benefits. The pace of development of transports using the WebRTC data channel, and the rapid deployment of the QUIC transport protocol, can both be attributed to using the combination of UDP as a substrate while providing confidentiality and authentication of the encapsulated transport headers and payload.
This document has described some current practises, and the implications for some stake holders, when transport layer header encryption is used. It does not judge whether these practises are necessary, or endorse the use of any specific practise. Rather, the intent is to highlight operational tools and practises to consider when designing and modifying transport protocols, so protocol designers can make informed choice about what transport header fields to encrypt, and whether it might be beneficial to make an explicit choice to expose certain fields to the network. In making such a decision, it is important to balance:
Observable transport header information might be useful to various stake holders. Other sets of stake holders have incentives to limit what can be observed. This document does not make recommendations about what information ought to be exposed, to whom it ought to be observable, or how this will be achieved. There are also design choices about where observable fields are placed. For example, one location could be a part of the transport header outside of the encryption envelope, another alternative is to carry the information in a network-layer option or extension header. New transport protocol designs ought to explicitly identify any fields that are intended to be observed, consider if there are alternative ways of providing the information, and reflect on the implications of observable fields being used by network devices, and how this might impact user privacy and protocol evolution when these fields become ossified.
As [RFC7258] notes, "Making networks unmanageable to mitigate (pervasive monitoring) is not an acceptable outcome, but ignoring (pervasive monitoring) would go against the consensus documented here." Providing explicit information can help avoid traffic being inappropriately classified, impacting application performance. An appropriate balance will emerge over time as real instances of this tension are analysed [RFC7258]. This balance between information exposed and information hidden ought to be carefully considered when specifying new transport protocols.
This document is about design and deployment considerations for transport protocols. Issues relating to security are discussed throughout this document.
Authentication, confidentiality protection, and integrity protection are identified as Transport Features by [RFC8095]. As currently deployed in the Internet, these features are generally provided by a protocol or layer on top of the transport protocol [I-D.ietf-taps-transport-security].
Confidentiality and strong integrity checks have properties that can also be incorporated into the design of a transport protocol or to modify an existing transport. Integrity checks can protect an endpoint from undetected modification of protocol fields by network devices, whereas encryption and obfuscation or greasing can further prevent these headers being utilised by network devices. Preventing observation of headers provides an opportunity for greater freedom to update the protocols and can ease experimentation with new techniques and their final deployment in endpoints. A protocol specification needs to weigh the costs of ossifying common headers, versus the potential benefits of exposing specific information that could be observed along the network path to provide tools to manage new variants of protocols.
Header encryption can provide confidentiality of some or all of the transport header information. This prevents an on-path device from knowledge of the header field. It therefore prevents mechanisms being built that directly rely on the information or seeks to infer semantics of an exposed header field. Reduces visibility into transport metadata can limit the ability to measure and characterise traffic. It can also provide privacy benefits in some cases.
Extending the transport payload security context to also include the transport protocol header protects both information with the same key. A privacy concern would arise if this key was shared with a third party, e.g., providing access to transport header information to debug a performance issue, would also result in exposing the transport payload data to the same third party. Such risks would be mitigated using a layered security design that provides one domain of protection and associated keys for the transport payload and encrypted transport headers; and a separate domain of protection and associated keys for any observable transport header fields.
Exposed transport headers are sometimes utilised as a part of the information to detect anomalies in network traffic. "While PM is an attack, other forms of monitoring that might fit the definition of PM can be beneficial and not part of any attack, e.g., network management functions monitor packets or flows and anti-spam mechanisms need to see mail message content." [RFC7258]. This can be used as the first line of defence to identify potential threats from DoS or malware and redirect suspect traffic to dedicated nodes responsible for DoS analysis, malware detection, or to perform packet "scrubbing" (the normalisation of packets so that there are no ambiguities in interpretation by the ultimate destination of the packet). These techniques are currently used by some operators to also defend from distributed DoS attacks.
Exposed transport header fields can also form a part of the information used by the receiver of a transport protocol to protect the transport layer from data injection by an attacker. In evaluating this use of exposed header information, it is important to consider whether it introduces a significant DoS threat. For example, an attacker could construct a DoS attack by sending packets with a sequence number that falls within the currently accepted range of sequence numbers at the receiving endpoint, this would then introduce additional work at the receiving endpoint, even though the data in the attacking packet might not finally be delivered by the transport layer. This is sometimes known as a “shadowing attack”. An attack can, for example, disrupt receiver processing, trigger loss and retransmission, or make a receiving endpoint perform unproductive decryption of packets that cannot be successfully decrypted (forcing a receiver to commit decryption resources, or to update and then restore protocol state).
One mitigation to off-path attack is to deny knowledge of what header information is accepted by a receiver or obfuscate the accepted header information, e.g., setting a non-predictable initial value for a sequence number during a protocol handshake, as in [RFC3550] and [RFC6056], or a port value that can not be predicted (see Section 5.1 of [RFC8085]). A receiver could also require additional information to be used as a part of a validation check before accepting packets at the transport layer (e.g., utilising a part of the sequence number space that is encrypted; or by verifying an encrypted token not visible to an attacker). This would also mitigate against on-path attacks. An additional processing cost can be incurred when decryption has to be attempted before a receiver is able to discard injected packets.
Open standards motivate a desire for this evaluation to include independent observation and evaluation of performance data, which in turn suggests control over where and when measurement samples are collected. This requires consideration of the appropriate balance between encrypting all and no transport header information. Open data, and accessibility to tools that can help understand trends in application deployment, network traffic and usage patterns can all contribute to understanding security challenges.
The Security and Privacy Considerations in the Framework for Large-Scale Measurement of Broadband Performance (LMAP) [RFC7594] contain considerations for Active and Passive measurement techniques and supporting material on measurement context.
Addition of observable transport information to the path increases the information available to an observer and may, when this information can be linked to a node or user, reduce the privacy of the user. See the security considerations of [RFC8558].
XX RFC ED - PLEASE REMOVE THIS SECTION XXX
This memo includes no request to IANA.
The authors would like to thank Mohamed Boucadair, Spencer Dawkins, Tom Herbert, Jana Iyengar, Mirja Kuehlewind, Kyle Rose, Kathleen Moriarty, Al Morton, Chris Seal, Joe Touch, Brian Trammell, Chris Wood, Thomas Fossati, Martin Thomson, David Black, and other members of TSVWG for their comments and feedback.
This work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 688421, and the EU Stand ICT Call 4. The opinions expressed and arguments employed reflect only the authors' view. The European Commission is not responsible for any use that might be made of that information.
This work has received funding from the UK Engineering and Physical Sciences Research Council under grant EP/R04144X/1.
-00 This is an individual draft for the IETF community.
-01 This draft was a result of walking away from the text for a few days and then reorganising the content.
-02 This draft fixes textual errors.
-03 This draft follows feedback from people reading this draft.
-04 This adds an additional contributor and includes significant reworking to ready this for review by the wider IETF community Colin Perkins joined the author list.
Comments from the community are welcome on the text and recommendations.
-05 Corrections received and helpful inputs from Mohamed Boucadair.
-06 Updated following comments from Stephen Farrell, and feedback via email. Added a draft conclusion section to sketch some strawman scenarios that could emerge.
-07 Updated following comments from Al Morton, Chris Seal, and other feedback via email.
-08 Updated to address comments sent to the TSVWG mailing list by Kathleen Moriarty (on 08/05/2018 and 17/05/2018), Joe Touch on 11/05/2018, and Spencer Dawkins.
-09 Updated security considerations.
-10 Updated references, split the Introduction, and added a paragraph giving some examples of why ossification has been an issue.
-01 This resolved some reference issues. Updated section on observation by devices on the path.
-02 Comments received from Kyle Rose, Spencer Dawkins and Tom Herbert. The network-layer information has also been re-organised after comments at IETF-103.
-03 Added a section on header compression and rewriting of sections referring to RTP transport. This version contains author editorial work and removed duplicate section.
-04 Revised following SecDir Review
-05 Editorial pass and minor corrections noted on TSVWG list.
-06 Updated conclusions and minor corrections. Responded to request to add OAM discussion to Section 6.1.
-07 Addressed feedback from Ruediger and Thomas.
Section 2 deserved some work to make it easier to read and avoid repetition. This edit finally gets to this, and eliminates some duplication. This also moves some of the material from section 2 to reform a clearer conclusion. The scope remains focussed on the usage of transport headers and the implications of encryption - not on proposals for new techniques/specifications to be developed.
-08 Addressed feedback and completed editorial work, including updating the text referring to RFC7872, in preparation for a WGLC.
-09 Updated following WGLC. In particular, thanks to Joe Touch (specific comments and commentary on style and tone); Dimitri Tikonov (editorial); Christian Huitema (various); David Black (various). Amended privacy considerations based on SECDIR review. Emile Stephan (inputs on operations measurement); Various others.
Added summary text and refs to key sections. Note to editors: The section numbers are hard-linked.
-10 Updated following additional feedback from 1st WGLC. Comments from David Black; Tommy Pauly; Ian Swett; Mirja Kuehlewind; Peter Gutmann; Ekr; and many others via the TSVWG list. Some people thought that "needed" and "need" could represent requirements in the document, etc. this has been clarified.
-11 Updated following additional feedback from Martin Thomson, and corrections from other reviewers.
-12 Updated following additional feedback from reviewers.
-13 Updated following 2nd WGLC with comments from D.L.Black; T. Herbert; Ekr; and other reviewers.
-14 Update to resolve feedback to rev -13. This moves the general discussion of adding fields to transport packets to section 6, and discusses with reference to material in RFC8558.
-15 Feedback from D.L. Black, T. Herbert, J. Touch, S. Dawkins and M. Duke. Update to add reference to RFC7605. Clarify a focus on immutable transport fields, rather than modifying middleboxes with Tom H. Clarified Header Compression discusion only provides a list of examples of HC methods for transport. Clarified port usage with Tom H/Joe T. Removed some duplicated sentences, and minor edits. Added NULL-ESP. Improved after initial feedback from Martin Duke.