RATS J. Zhang Internet-Draft H. Labiod Intended status: Standards Track Huawei Technologies France S.A.S.U. Expires: 24 April 2025 T. Li Shield Lab, Singapore Research Center, Huawei Technologies T. Giannetsos UBITECH Ltd. H. Birkholz Fraunhofer SIT 21 October 2024 Handling Multiple Verifiers in the RATS Architecture draft-zhang-rats-multiverifiers-01 Abstract In the IETF Remote Attestation Procedures (RATS) architecture, a Verifier accepts Evidence and generates Attestation Results needed by Relying Parties. This document provides a solution to inconsistent behaviors of the Verifier in the RATS architecture by introducing a mechanism to aggregate Attestation Results collected from multiple Verifiers at the Relying Party while simplifying its policy and operation. Status of This Memo 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 24 April 2025. Copyright Notice Copyright (c) 2024 IETF Trust and the persons identified as the document authors. All rights reserved. Zhang, et al. Expires 24 April 2025 [Page 1] Internet-Draft Multiple RATS Verifiers October 2024 This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/ license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 1.1. Passport Model Cases . . . . . . . . . . . . . . . . . . 3 1.2. Background-check Model Cases . . . . . . . . . . . . . . 4 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 6 3. Handing Multiple Verifiers . . . . . . . . . . . . . . . . . 6 3.1. Aggregation of Attestation Results from Multiple Verifiers . . . . . . . . . . . . . . . . . . . . . . . . 7 3.2. Verifier Manager . . . . . . . . . . . . . . . . . . . . 8 4. Use Cases . . . . . . . . . . . . . . . . . . . . . . . . . . 10 5. Security Consideration . . . . . . . . . . . . . . . . . . . 10 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 10 7.1. Normative References . . . . . . . . . . . . . . . . . . 11 7.2. Informative References . . . . . . . . . . . . . . . . . 11 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 11 1. Introduction The Remote Attestation procedures (RATS) Architecture illustrates an overview of the roles and data-flows between them in Section 3 of [RFC9334]. Section 5 of [RFC9334] refines the data-flow diagram by describing two reference models: Passport Model and Background- check Model. As discussed in that document, a Verifier accepts Evidence from Attesters, appraises it using Appraisal Policy, and generates Attestation Results needed by Relying Parties. As a single Verifier can introduce a single point of failure, either as the target of a denial of service attack, due to compromization, service congestion, or broken Internet connectivity to the Verifier, relying on a single trusted entity can introduce significant risk. The architectural pattern of using multiple Verifiers are one approach to counter such risks. Nevertheless, it is not guaranteed that different Verifiers generate the same Attestation Results. Some exemplary reasons include: a) RATS conceptual messages, such as Reference Values, Endorsements, Appraisal Policy for Evidence for different Verifiers, are not necessarily aligned, b) certain Zhang, et al. Expires 24 April 2025 [Page 2] Internet-Draft Multiple RATS Verifiers October 2024 Verifiers can be compromised, or c) some Verifiers follow different Appraisal Policies for Evidence. This lack of alignment can result in significant issues in both Passport Model and Background-check Model, which is detailed as follows. The Solution to address the problem of the lack of alignment is detailed in Section 3. 1.1. Passport Model Cases Under the Passport Model, an Attester sends Evidence to a Verifier. The Verifier generates the Attestation Results and sends these back to the Attester. The Attester conveys the Attestation Results to the Relying Party to proof its trustworthiness. Fig. 1 and 2 show scenarios that multiple heterogeneous Verifiers can introduce issues in a Passport Model based system. In Fig. 1, if Verifier A is not trusted by the Relying Party, Attestation Results sent by the Attester can always be rejected by the Relying Party, which means that the Attester may end up in a loop of producing and conveying Attestation Evidence and wait for Attestation Results in vain, repeatedly. In Fig. 2, Verifier A generates positive Attestation Results for an Attester, while Verifier B generates negative Attestation Results for the same Attester. To trick a Relying Party into putting unjustified trust in the Attester, an Attester can act maliciously by selectively forwarding only Attestation Results from Verifier A and not Verifier B. Such malicious behavior would render a trustworthiness assessment of Attesters by the Relying Party biased or unreliable. .-------------. | | Compare Evidence | Verifier A | against appraisal policy | | '--------+----' ^ | Evidence | | Attestation | | Result | v .---+--------. .-------------. Compare | +------------>X| | Attestation | Attester | Attestation | Relying | Result against | | Result | Party | appraisal '------------' '-------------' policy Figure 1: Passport Model with Verifier A not trusted by Relying Party. Zhang, et al. Expires 24 April 2025 [Page 3] Internet-Draft Multiple RATS Verifiers October 2024 .-------------. | | Compare Evidence | Verifier A | against appraisal policy | | '--------+----' ^ | Evidence | | Attestation | | Result A (positive) | v .---+--------. .-------------. Compare | +------------->| | Attestation | Attester | Attestation | Relying | Result against | | Result A | Party | appraisal '---+--------' '-------------' policy | ^ Evidence | | Attestation | | Result B (negative) | | V | .--------+----. | | Compare Evidence | Verifier B | against appraisal policy | | '-------------' Figure 2: Passport Model with a cheating Attester 1.2. Background-check Model Cases Under the Background-check Model, an Attester sends Evidence to a Verifier via a Relaying Party, and the Verifier generates the Attestation Results and sends them back to the Relying Party. Fig. 3 and 4 show scenarios where multiple heterogeneous Verifiers introduce potential issues in a Background-check Model. In Fig. 3, even if a Verifier is trusted by a Relying Party, there is no assurance that it is working as intended and only does what it is supposed to do and nothing else. If multiple Verifiers exist, neither Evidence might reach all Verifiers nor all Attestation Results might reach the Relying Party due to failing conveyance mechanisms, or due to the Verifier itself being compromised or malfunctioning., or hardware problems. Zhang, et al. Expires 24 April 2025 [Page 4] Internet-Draft Multiple RATS Verifiers October 2024 In Fig. 4, a Relying Party is able to alternate between Verifiers. When these Verifiers are heterogeneous though, a Relying Party might receive different or conflicting Attestation Results from them, which means the trustworthy assessment of the Attester can rely (and fail) on a specific selection of Verifiers made by at the Relying Party side. .-------------. | | Compare Evidence | Verifier | against | | appraisal | x(2) | policy '--------+----' ^ x(3) Evidence | | Attestation x(1)| Result | v .------------. .----|--------. | +-------------->|---' | Compare | | | | Attestation | Attester | Evidence | Relying | Result against | | | Party | appraisal policy '------------' '-------------' Figure 3: A Background-Check Model where a Verifier is not available because of 1) a Relying Party not being reachable by the Verifier, 2) a malfunction of the Verifier. Zhang, et al. Expires 24 April 2025 [Page 5] Internet-Draft Multiple RATS Verifiers October 2024 .-------------. | | Compare Evidence | Verifier | against | A | appraisal '--------+----' policy ^ | Evidence | | Attestation | | Result (positive) | v .------------. .----|--------. Compare | +-------------->|---' | Attestation | Attester | Evidence | Relying | Result against | | | Party | appraisal policy '------------' '----+--------' | ^ Evidence | | Attestation | | Result (negative) v | .--------+----. | | Compare Evidence | Verifier | against | B | appraisal '-------------' policy Figure 4: A Background-Check Model conveying conflicting Attestation Results originating from multiple Verifiers. 2. Terminology The following terms are imported from [RFC9334]: Attester, Evidence, Endorsement, Reference value, Appraisal Policy, Relying Party, and Verifier. Also imported are the time definitions time(VG), time(NS), time(EG), time(ER), time(RG),time(RX), and time(OP) from that document's Appendix A. New relevant Events over Time: time(AG): the time at the event that the Attestation Results for the same attester is aggregated. 3. Handing Multiple Verifiers In this section, we show the data-flow to support robust aggregation of the Attestation Results in an in an environment with many Verifiers that may be heterogeneous. Here heterogeneous means that the Verifiers may generate different Attestation Results according to the same input of the Evidence. Zhang, et al. Expires 24 April 2025 [Page 6] Internet-Draft Multiple RATS Verifiers October 2024 Below are the examples that the Relying Party needs multiple Verifiers. 1) To get Attestation Results from multiple Verifiers that follow the same golden measurement, to provide resillience against failure or compromisation of certain Verifiers. In this case, the Attestation Results are expected to be the same from these Verifiers. 2) Different Verifiers provide different Attestation Results according to different sub part of the same Evidence. The Relying Party makes it own judgement according to its own logic to combine these heterogeneous Attestation Results. In this case, the Attestation Results can be expected to be different from different Verifiers. In terms of resilience, these Verifiers cannot be replaced by a conceptual single (proxy) Verifier as this single Verifier may still has the availability issue. As an example, we extend the attestation data-flow based on the Background-Check Model to handel multiple Verifiers that guarantee the freshness of the Evidence from the Attester. The Verifier Manager is introduced into the attestation system to help the Relying Party choosing Verifiers that are aggregable according to its own logic, that is, the Relying Party has one mechanism to combine Attestation Results from these Verifiers to make the final conclusion. 3.1. Aggregation of Attestation Results from Multiple Verifiers Fig. 5 below is a sequence diagram which updates Fig. 14 in [RFC9334] to support the aggregation of Attestation Results from multiple Verifiers in a Background-check Model. The nonce is generated by the Relying Party, in place of each Verifier, so as to reduce the amount of Evidence generated. The aggregation method implemented by the Relying Party is out of scope of this draft. For example, the majority vote could be viewed as a possible solution, when the Verifiers are expected to follow the same standard. The Attestation Results can be aggregated to help the Relying Party making the decision, or be aggregated as a new Attestation Result, which is decided by the Relying Party itself. Zhang, et al. Expires 24 April 2025 [Page 7] Internet-Draft Multiple RATS Verifiers October 2024 .---------. .--------. .--------. .--------. .---. | Attester| |Verifier| |Verifier| |Verifier| | RP| | | | 1 | | 2 | | k | | | '---------' '--------' '--------' '--------' '---' | | | | | Time(VG_a) ~ ~ ~ ~ | | | | | |<----Nonce---------------------------------------time(NS_r) Time(EG_a) | | | | | | | | | |-----Evidence{Nonce}----------------------------------->| | | time(ER_r_1) | |<-----Evidence{Nonce}---------------------| | | | | time(ER_r_2) | time(RG_v_1) |<-Evidence{Nonce}---------------| | | time(RG_v_2) | time(ER_r_k) | | | |<-Evidence{Nonce}--| | | | time(RG_v_k) | | |--Attestation Result--------------------->| | | {time(RX_v_1)-time(RG_v_1)} | | | |----Attestation Result--------->| | | | {time(RX_v_2)-time(RG_v_2)} | | | | |--------AR------->| | | | {time(RX_v_k)-time(RG_v_k)} | | | | | time(AG_r) | | | | time(OP_r) Figure 5: Background-Check Model with the support of the aggregation of Attestation Results from multiple Verifiers. 3.2. Verifier Manager Manually configuring the Verifiers in each Relying Party is not well adapted to the changing of the network environment. As there is no guarantee of the availability and consolidation of these Verifiers in the long term, we introduce a new entity in RATS architecture, which is the Verifier Manager, to address these issues. As shown in Fig. 6, after configuring the anchor seed Verifiers in the Relying Party, which is typically a small set of trusted Verifiers by the Relying Party. The Relying Party can communicate with the Verifier Manager with this list of Verifiers, in together with certain parameters. The Verifier manager matches this list with its local database of the groups of Verifiers, find Verifiers that matches the parameter. Then the Verifier Manager sends these Verifiers back to the Relying Party, as its recommended Verifiers list. In such a way, each Relying Party can flexibly configure its policy for the trusted Verifier, without knowing the detail of every Verifier. At the Verifier Manager side, the Verifiers in the same group are expected to follow the same Zhang, et al. Expires 24 April 2025 [Page 8] Internet-Draft Multiple RATS Verifiers October 2024 golden measurement, that is, they are expected to generate the same Attestation Results when they receive the same Evidences. The example are the Verifiers that are deployed by the same company or the alliance. Here the same for Evidences and Attestation Results are in the sense of semantic, that is, they can be wrapped in different formats, CWT or JWT for example, but the content itself is the same. When a Relying Party receives certain minority attesation results from certain Verifiers, it can inform the Verifier Manager this incidence, and the Verifier Manager will reduce the reputation of these verifiers, and reduce the probability to recommend these Verifiers to Relying Parties. So in the long run, the misbehaved Verifiers will be punished. The details on the reputation management scheme for Verifiers are out of scope of this draft. .---------. .----------. .----------. .--------------. | Endorser| | Reference | | Verifier | | Relying Party| '+--------' | Value | | Owner | | Owner | | | Provider | '----+-----' '-----+--------' | '------+----' | | | | | | | Endorsements | Reference | Appraisal | Appraisal | | Values | Policy for | Policy for | | | Evidence | Attestation '-----------. | | | Results | | | | v v v | .-------------------------. | .------>| Verifier +------. | | '-------------------------' | | | | | | Evidence Attestation | | | Results | | | | | | v v .-----+----. .---------------. | Attester | | Relying Party | '----------' '---------------' | ^ Anchor seed Verifiers, | | Recommended parameter | | Verifiers | | .------------------. | Verifier Manager | '------------------' Figure 6: Revised Data Flow based on RFC9334 Zhang, et al. Expires 24 April 2025 [Page 9] Internet-Draft Multiple RATS Verifiers October 2024 4. Use Cases This Section illustrates some use cases that can benefit from an architecture that takes multiple Verifiers into account. Use case 1: Node Attestation for Trusted Routing Need: Trustworthiness Assessment of routing nodes (Attesters) against Verifier while ensuring the robustiness of the attestation verification service (AVS) Solution: Provide multiple Verifiers (primary and secondaries) to ensure the availability of the AVS for nodes in the network Source: Trusted Path Routine [I-D.voit-rats-trustworthy-path-routing], network attestation for secure routing [I-D.liu-nasr-requirements] Use case 2: Intent-driven Attestation Classification for Data Center Network Solutions Need: In Data Centers, Data Processing Units (DPU) need to attest other units (DPUs, CPUs, GPUs) to determine their states. There might be hundreds of Verifiers (DPUs) for one Attester (DPU/CPU/GPU). At the Attester side, to generate indididually one Evidence for each Verifier could be prohibitive. Solution: One Verifier works as the Relying Party to contact the Attester, and marks other Verifiers that need to attest this Attester in the same interesting group. sends the attestation request to the Attester. The Evidence from the Attester is only generated once and sent to this Verifier. This Verifier forwards the Evidence to other Verifiers that in the same interesting group and obtain the Attestation Results from them. It generates the Aggregated Attestation Results and shares it within the Verifiers in the same interesting group. In this manner, the Attester does not need to generate the Evidence for every Verifier, and the attestation procedure works even when certain Verifier does not work. 5. Security Consideration [TBD] 6. IANA Considerations [TBD] 7. References Zhang, et al. Expires 24 April 2025 [Page 10] Internet-Draft Multiple RATS Verifiers October 2024 7.1. Normative References [I-D.voit-rats-trustworthy-path-routing] Birkholz, H., Voit, E., Liu, P. C., Lopez, D., and M. Chen, "Trusted Path Routing", Work in Progress, Internet- Draft, draft-voit-rats-trustworthy-path-routing-10, 8 July 2024, . [RFC9334] Birkholz, H., Thaler, D., Richardson, M., Smith, N., and W. Pan, "Remote ATtestation procedureS (RATS) Architecture", RFC 9334, DOI 10.17487/RFC9334, January 2023, . 7.2. Informative References [I-D.liu-nasr-requirements] Liu, P. C., Iannone, L., Lopez, D., Pastor, A., Chen, M., and L. Su, "NASR Use Case and Requirements", Work in Progress, Internet-Draft, draft-liu-nasr-requirements-03, 20 October 2024, . [RFC9397] Pei, M., Tschofenig, H., Thaler, D., and D. Wheeler, "Trusted Execution Environment Provisioning (TEEP) Architecture", RFC 9397, DOI 10.17487/RFC9397, July 2023, . [TAP] Trusted Computing Group, "TCG Trusted Attestation Protocol (TAP) Use Cases for TPM Families 1.2 and 2.0 and DICE", 1.0 , 5 November 2019, . Authors' Addresses Jun Zhang Huawei Technologies France S.A.S.U. 18, Quai du Point du Jour 92100 Boulogne-Billancourt France Email: junzhang1@huawei.com Zhang, et al. Expires 24 April 2025 [Page 11] Internet-Draft Multiple RATS Verifiers October 2024 Houda Labiod Huawei Technologies France S.A.S.U. 18, Quai du Point du Jour 92100 Boulogne-Billancourt France Email: houda.labiod@huawei.com Tieyan Li Shield Lab, Singapore Research Center, Huawei Technologies Science Park II., 20 Science Park Road, SINGAPORE Teletech Park Singapore Email: Li.Tieyan@huawei.com Thanassis Giannetsos UBITECH Ltd. Thessalias 8 and Etolias 10 GR-15231 Chalandri, Greece Email: agiannetsos@ubitech.eu Henk Birkholz Fraunhofer SIT Rheinstrasse 75 64295 Darmstadt Germany Email: henk.birkholz@sit.fraunhofer.de Zhang, et al. Expires 24 April 2025 [Page 12]