Internet Engineering Task Force (IETF) T. Hardie Request for Comments: 8165 May 2017 Category: Informational ISSN: 2070-1721
Design Considerations for Metadata Insertion
Abstract
The IAB published RFC 7624 in response to several revelations of pervasive attacks on Internet communications. This document considers the implications of protocol designs that associate metadata with encrypted flows. In particular, it asserts that designs that share metadata only by explicit actions at the host are preferable to designs in which middleboxes insert metadata.
Status of This Memo
This document is not an Internet Standards Track specification; it is published for informational purposes.
This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Not all documents approved by the IESG are a candidate for any level of Internet Standard; see Section 2 of RFC 7841.
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc8165.
Copyright Notice
Copyright (c) 2017 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 (http://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.
Hardie Informational [Page 1]
RFC 8165 Design Considerations for Metadata Insertion May 2017
To minimize the risks associated with pervasive surveillance, it is necessary for the Internet technical community to address the vulnerabilities exploited in the attacks documented in [RFC7258] and the threats described in [RFC7624]. The goal of this document is to address a common design pattern that emerges from the increase in encryption: explicit association of metadata that would previously have been inferred from the plaintext protocol.
This document makes extensive use of standard security and privacy terminology; see [RFC4949] and [RFC6973]. Readers should be familiar with the terms defined in [RFC6973], including "Eavesdropper", "Observer", "Initiator", "Intermediary", "Recipient", "Attack" (in a privacy context), "Correlation", "Fingerprint", "Traffic Analysis", and "Identifiability" (and related terms). Readers should also be familiar with terms that are specific to the attacks discussed in [RFC7624], including "Pervasive Attack", "Passive Pervasive Attack", "Active Pervasive Attack", "Observation", "Inference", and "Collaborator".
One of the core mitigations for the loss of confidentiality in the presence of pervasive surveillance is data minimization, which limits the amount of data disclosed to those elements absolutely required to complete the relevant protocol exchange. When data minimization is in effect, some information that was previously available may be removed from specific protocol exchanges. The information may be removed explicitly (for example, by a browser suppressing cookies
Hardie Informational [Page 2]
RFC 8165 Design Considerations for Metadata Insertion May 2017
during private modes) or by other means. As noted in [RFC7624], some topologies that aggregate or alter the network path also act to reduce the ease with which metadata is available to eavesdroppers.
In some cases, other actors within a protocol context will continue to have access to the information that has been thus withdrawn from specific protocol exchanges. If those actors attach the information as metadata to those protocol exchanges, the confidentiality effect of data minimization is lost.
Restoring information is particularly tempting at systems not primarily deployed to increase confidentiality. A proxy providing compression, for example, may wish to restore the identity of the requesting party; similarly, a VPN system used to provide channel security may believe that the origin IP should be restored. Actors considering restoring metadata may believe that they understand the relevant privacy considerations or believe that, because the primary purpose of the service was not privacy-related, none exist. Examples of this design pattern include [RFC7239] and [RFC7871].
Avoid inserting metadata to restore information that would otherwise be unavailable to later participants in a protocol exchange. It contributes to the overall loss of confidentiality for the Internet and trust in the Internet as a medium. Do not add metadata to flows at intermediary devices unless a positive affirmation of approval for restoration has been received from the actor whose data will be added.
Instead, design the protocol so that the actor can add such metadata themselves so that it flows end to end, rather than requiring the action of other parties. In addition to improving privacy, this approach ensures consistent availability between the communicating parties, no matter what path is taken. (Note that this document does not attempt to describe how an actor sets policies on providing this metadata, as the range of systems that might be implied is very broad).
As an example, RFC 7871 describes a method that had already been deployed and notes that it is unlikely that a clean-slate design would result in this mechanism. If a clean-slate design were built to follow the advice in this document, that design would likely not use a core element of RFC 7871: rather than adding metadata at a proxy, it would provide facilities for end systems to add it to their initial queries. In the case of RFC 7871, the relevant metadata is relatively easy for an end system to derive, as Session Traversal Utilities for NAT (STUN) [RFC5389] provides a method for learning the
Hardie Informational [Page 3]
RFC 8165 Design Considerations for Metadata Insertion May 2017
reflexive transport address from which a client subnet could be derived. This would allow clients to populate this data themselves, thus affirming their consent and providing data at a granularity with which they were comfortable. As in RFC 7871, the addition of this data would require confirmation that the upstream DNS resolver understands what to do with it, but the same negotiation mechanism, an Extension Mechanisms for DNS (EDNS(0)) option [RFC6891], could be used. Because of this negotiation, there would be a new variability in responses that would change the caching behavior for data supplied by participating servers. This is not a major change from the current design, however, as the same considerations set out in Sections 7.3.2 and 7.5 of RFC 7871 would apply to client-supplied subnets as well as to proxy-supplied subnets.
From a protocol perspective, in other words, this approach would be a minor change from RFC 7871, would be as fully featured, and would provide better privacy properties than the on-path update mechanism RFC 7871 provides. The next section examines why, despite this, deployment considerations have sometimes trumped cleaner designs.
There are a few common tensions associated with the deployment of systems that restore metadata. The first is the trade-off in speed of deployment for different actors. The Forwarded HTTP Extension in [RFC7239] provides an example of this. When used with a proxy, it restores information related to the original requesting party, thus allowing a responding server to tailor responses according to the original party's region, network, or other characteristics associated with the identity. It would, of course, be possible for the originating client to add this data itself, after using STUN [RFC5389] or a similar mechanism to first determine the information to declare. This would require, however, full specification and adoption of this mechanism by the end systems. It would not be available at all during this period and would thereafter be limited to systems that have been upgraded to include it. The long tail of browser deployments indicates that many systems might go without upgrades for a significant period of time. The proxy infrastructure, in contrast, is commonly under more active management and represents a much smaller number of elements; this impacts both the general deployment difficulty and the number of systems that the origin server must trust.
The second common tension is between metadata minimization and the desire to tailor content responses. For origin servers whose content is common across users, the loss of metadata may have limited impact on the system's functioning. For other systems, which commonly tailor content by region or network, the loss of metadata may imply a
Hardie Informational [Page 4]
RFC 8165 Design Considerations for Metadata Insertion May 2017
loss of functionality. Where the user desires this functionality, restoration can commonly be achieved by the use of other identifiers or login procedures. Where the user does not desire this functionality, but it is a preference of the server or a third party, adjustment is more difficult. At the extreme, content blocking by network origin may be a regulatory requirement. Trusting a network intermediary to provide accurate data is, of course, fragile in this case, but it may be a part of the regulatory framework.
There are also tensions with latency of operation. For example, where the end system does not initially know the information that would be added by on-path devices, it must engage the protocol mechanisms to determine it. Determining a public IP address to include in a locally supplied header might require a STUN exchange, and the additional latency of this exchange discourages deployment of host-based solutions. To minimize this latency, engaging those mechanisms may need to be done in parallel with or in advance of the core protocol exchanges with which this metadata would be supplied.
These tensions do not change the basic recommendation, but they suggest that the parties who are introducing encryption and data minimization for existing protocols consider carefully whether the work also implies introducing mechanisms for the end-to-end provisioning of metadata when a user has actively consented to provide it.
This memorandum describes a design pattern emerging from responses to the attacks described in [RFC7258]. Continued use of this design pattern, which uses mid-flow devices to restore metadata, lowers the impact of mitigations to that attack.
Note that some emergency service recipients, notably PSAPs (Public Safety Answering Points) may prefer data provided by a network to data provided by an end system, because an end system could use false data to attack others or consume resources. While this has the consequence that the data available to the PSAP is often more coarse than that available to the end system, the risk of false data being provided involves a risk to the lives of those targeted.
Hardie Informational [Page 5]
RFC 8165 Design Considerations for Metadata Insertion May 2017
[RFC7624] Barnes, R., Schneier, B., Jennings, C., Hardie, T., Trammell, B., Huitema, C., and D. Borkmann, "Confidentiality in the Face of Pervasive Surveillance: A Threat Model and Problem Statement", RFC 7624, DOI 10.17487/RFC7624, August 2015, <http://www.rfc-editor.org/info/rfc7624>.
This document is derived in part from the work initially done on the perpass mailing list and at the STRINT workshop [RFC7687]. The text was originally developed by the IAB's Privacy and Security Program before submission to the IETF. The document also benefited from an extensive review by Mohamed Boucadair.