4.2. Primary Client - Primary Server Negotiation Phase
The two entities exchange a set of data to ensure the respective identities.¶
They use HPKE KEM to negotiate a "SharedSecret".¶
| Internet-Draft | TODO - Abbreviation | March 2023 |
| Elkins, et al. | Expires 27 September 2023 | [Page] |
RFC8250 describes an optional Destination Option (DO) header embedded in each packet to provide sequence numbers and timing information as a basis for measurements. As this data is sent in clear-text, this may create an opportunity for malicious actors to get information for subsequent attacks. This document defines PDMv2 which has a lightweight handshake (registration procedure) and encryption to secure this data. Additional performance metrics which may be of use are also defined.¶
This note is to be removed before publishing as an RFC.¶
The latest revision of this draft can be found at https://ameyand.github.io/PDMv2/draft-elkins-ippm-encrypted-pdmv2.html. Status information for this document may be found at https://datatracker.ietf.org/doc/draft-ietf-ippm-encrypted-pdmv2/.¶
Discussion of this document takes place on the IP Performance Measurement Working Group mailing list (mailto:ippm@ietf.org), which is archived at https://mailarchive.ietf.org/arch/browse/ippm/. Subscribe at https://www.ietf.org/mailman/listinfo/ippm/.¶
Source for this draft and an issue tracker can be found at https://github.com/ameyand/PDMv2.¶
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 27 September 2023.¶
Copyright (c) 2023 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 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.¶
The current PDM is an IPv6 Destination Options header which provides information based on the metrics like Round-trip delay and Server delay. This information helps to measure the Quality of Service (QoS) and to assist in diagnostics. However, there are potential risks involved transmitting PDM data during a diagnostics session.¶
PDM metrics can help an attacker understand about the type of machine and its processing capabilities. Inferring from the PDM data, the attack can launch a timing attack. For example, if a cryptographic protocol is used, a timing attack may be launched against the keying material to obtain the secret.¶
Along with this, PDM does not provide integrity. It is possible for a Man-In-The-Middle (MITM) node to modify PDM headers leading to incorrect conclusions. For example, during the debugging process using PDM header, it can mislead the person showing there are no unusual server delays.¶
PDMv2 adds confidentiality, integrity and authentication to PDM.¶
PDMv2 consists of three kinds of flows:¶
These terms are defined in Section 3. Sample topologies may be found in Appendix 1.¶
This document describes the Secondary to Secondary protocol and security requirements. The Primary to Primary and Primary to Secondary protocol will be described in a subsequent document.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
Note: a client may act as a server (have listening ports).¶
The protocol will proceed in 3 steps.¶
After-the-fact (or real-time) data analysis of PDM flow may occur by network diagnosticians or network devices. The definition of how this is done is out of scope for this document.¶
Enterprises have many servers and many clients. These clients and servers may be in multiple locations. It may be less overhead to have a secure location (ex. Shared database) for servers and clients to share keys. Otherwise, each client needs to keep track of the keys for each server.¶
Please view Appendix 1 for some sample topologies and further explanation.¶
+-----------+ +-----------+
|Primary |<====================>|Primary |
|Client (PC)| |Server (PS)|
+-----+-----+ +-----+-----+
|| ||
|| ||
+-------------------------+ +-------------------------+
| Secondary Clients(SC's) | | Secondary Servers (SS's)|
| | | |
| +----+ +----+ +----+ | | +----+ +----+ +----+ |
| |SC1 | |SC2 |.. |SC N| |<=======>| |SS 1| |SS 2|.. |SS N| |
| +----+ +----+ +----+ | | +----+ +----+ +----+ |
| | | |
+-------------------------+ +-------------------------+
¶
The two entities exchange a set of data to ensure the respective identities.¶
They use HPKE KEM to negotiate a "SharedSecret".¶
The "SharedSecret" is shared securely:¶
Each Client and Server derive a "SessionTemporaryKey" by using HPKE KDF, using the following inputs:¶
The Kri SHOULD be initialized to zero.¶
The server and client initialize (separately) a pseudo-random non- repeating sequence between 1 and 2^15-1. How to generate this sequence is beyond the scope of this document, and does not affect the rest of the specification. When the sequence is used fully, or earlier if appropriate, the sender signals the other party that a key change is necessary. This is achieved by flipping the "F bit" and resetting the PRSEQ. The receiver increments the Kri of the sender, and derives another SessionTemporaryKey to be used for decryption.¶
It shall be stressed that the two SessionTemporaryKeys used in the communication are never the same, as the 5-tuple is reversed for the Server and Client. Moreover, the time evolution of the respective Kri can be different. As a consequence, each entity must maintain a table with (at least) the following informations:¶
An implementation might optimize this further by caching the OwnSessionTemporaryKey (used in Encryption) and OtherSessionTemporaryKey (used in Decryption).¶
As discussed in the introduction, PDM data can represent a serious data leakage in presence of a malicious actor.¶
In particular, the sequence numbers included in the PDM header allows correlating the traffic flows, and the timing data can highlight the operational limits of a server to a malicious actor. Moreover, forging PDM headers can lead to unnecessary, unwanted, or dangerous operational choices, e.g., to restore an apparently degraded Quality of Service (QoS).¶
Due to this, it is important that the confidentiality and integrity of the PDM headers is maintained. PDM headers can be encrypted and authenticated using the methods discussed in Section 5.4, thus ensuring confidentiality and integrity. However, if PDM is used in a scenario where the integrity and confidentiality is already ensured by other means, they can be transmitted without encryption or authentication. This includes, but is not limited to, the following cases:¶
PDM data must be kept confidential between the intended parties, which includes (but is not limited to) the two entities exchanging PDM data, and any legitimate party with the proper rights to access such data.¶
PDM data must not be forged or modified by a malicious entity. In other terms, a malicious entity must not be able to generate a valid PDM header impersonating an endpoint, and must not be able to modify a valid PDM header.¶
An unauthorized party must not be able to send PDM data and must not be able to authorize another entity to do so. The protocol to define this will be defined in a subsequent document. Alternatively, if authentication is done via any of the following, this requirement may be seen to be met.¶
Symmetric key cryptography has performance benefits over asymmetric cryptography; asymmetric cryptography is better for key management. Encryption schemes that unite both have been specified in [RFC1421], and have been participating practically since the early days of public-key cryptography. The basic mechanism is to encrypt the symmetric key with the public key by joining both yields. Hybrid public-key encryption schemes (HPKE) [RFC9180] used a different approach that generates the symmetric key and its encapsulation with the public key of the receiver.¶
Our choice is to use the HPKE framework that incorporates key encapsulation mechanism (KEM), key derivation function (KDF) and authenticated encryption with associated data (AEAD). These multiple schemes are more robust and significantly efficient than the traditional schemes and thus lead to our choice of this framework.¶
The IPv6 Destination Options extension header [RFC8200] is used to carry optional information that needs to be examined only by a packet's destination node(s). The Destination Options header is identified by a Next Header value of 60 in the immediately preceding header and is defined in RFC 8200 [RFC8200]. The IPv6 PDMv2 destination option is implemented as an IPv6 Option carried in the Destination Options header.¶
The IPv6 PDMv2 destination option contains the following base fields:¶
PDMv2 adds a new metric to the existing PDM [RFC8250] called the Global Pointer. The existing PDM fields are identified with respect to the identifying information called a "5-tuple".¶
The 5-tuple consists of:¶
Unlike PDM fields, Global Pointer (GLOBALPTR) field in PDMv2 is defined for the SADDR type. Following are the SADDR address types considered:¶
The Global Pointer is treated as a common entity over all the 5-tuples with the same SADDR type. It is initialised to the value 1 and increments for every packet sent. Global Pointer provides a measure of the amount of IPv6 traffic sent by the PDMv2 node.¶
When the SADDR type is Link-Local, the PDMv2 node sends Global Pointer defined for Link-Local addresses, and when the SADDR type is Global Unicast, it sends the one defined for Global Unicast addresses.¶
PDMv2 has two different header formats corresponding to whether the metric contents are encrypted or unencrypted. The difference between the two types of headers is determined from the Options Length value.¶
Following is the representation of the unencrypted PDMv2 header:¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Option Type | Option Length | Vrsn | Reserved Bits | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Random Number |f| ScaleDTLR | ScaleDTLS | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Global Pointer | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | PSN This Packet | PSN Last Received | |-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Delta Time Last Received | Delta Time Last Sent | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Following is the representation of the encrypted PDMv2 header:¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Option Type | Option Length | Vrsn | Reserved Bits | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Random Number |f| | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ : | Encrypted PDM Data : : (30 bytes) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Option Type¶
0x0F¶
8-bit unsigned integer. The Option Type is adopted from RFC 8250 [RFC8250].¶
Option Length¶
0x12: Unencrypted PDM¶
0x22: Encrypted PDM¶
8-bit unsigned integer. Length of the option, in octets, excluding the Option Type and Option Length fields. The options length is used for differentiating PDM [RFC8250], unencrypted PDMv2 and encrypted PDMv2.¶
Version Number¶
0x2¶
4-bit unsigned number.¶
Reserved Bits¶
12-bits.¶
Reserved bits for future use. They are initialised to 0 for PDMv2.¶
Random Number¶
15-bit unsigned number.¶
This is a random number with as much entropy as desired by the implementation. The level of entropy should be clearly specified to the user.¶
Flag Bit¶
1-bit field.¶
The flag bit indicates that the sender has used a new SessionTemporaryKey and the receiver should increment the Kri of the sender and derive the same new SessionTemporaryKey.¶
Scale Delta Time Last Received (SCALEDTLR)¶
8-bit unsigned number.¶
This is the scaling value for the Delta Time Last Sent (DELTATLS) field.¶
Scale Delta Time Last Sent (SCALEDTLS)¶
8-bit unsigned number.¶
This is the scaling value for the Delta Time Last Sent (DELTATLS) field.¶
Global Pointer¶
32-bit unsigned number.¶
Global Pointer is initialized to 1 for the different source address types and incremented monotonically for each packet with the corresponding source address type.¶
This field stores the Global Pointer type corresponding to the SADDR type of the packet.¶
Packet Sequence Number This Packet (PSNTP)¶
16-bit unsigned number.¶
This field is initialized at a random number and is incremented monotonically for each packet of the 5-tuple.¶
Packet Sequence Number Last Received (PSNLR)¶
16-bit unsigned number.¶
This field is the PSNTP of the last received packet on the 5-tuple.¶
Delta Time Last Received (DELTATLR)¶
16-bit unsigned integer.¶
The value is set according to the scale in SCALEDTLR.¶
Delta Time Last Received = (send time packet n - receive time packet (n - 1))¶
Delta Time Last Sent (DELTATLS)¶
16-bit unsigned integer.¶
The value is set according to the scale in SCALEDTLS.¶
Delta Time Last Sent = (receive time packet n - send time packet (n - 1))¶
PDMv2 DOH can be used by an attacker to gather information about a victim (passive attack) or to force the victim to modify its operational parameters to comply with forged data (active attacks).¶
In order to mitigate these, it is important that the PDMv2 DOH is subject to:¶
with respect to an attacker.¶
In the following we will refer to two different "groups", that can or cannot belong to the same operational and management domain:¶
We will assume, for the sake of generalization, that the Servers are managed by an Organization (OrgA) implementing management procedures over them, and the Clients by a different Organization (OrgB).¶
An attacker could be in the following positions:¶
Furthermore, since PDMv2 DOH encryption could consume resources (albeit limited), it is possible to foresee a call of DoS by resource exhaustion. Hence, it is relevant to consider a form of access control to verify that the Server and Client belong to OrgA and OrgB respectively. This could be a delegated trust.¶
In other terms, a Client could just want to verify that the Server belongs to OrgA, without actually verifying the identity of the Server.¶
The Authentication and Authorization of Clients and Servers is thus delegated to the respective Organizations. In other terms, we do not expect, or want, that a Client and a Server should be forced to verify the respective identities (Authentication) or the permissions to use PDMv2 (Authorization).¶
The simple knowledge of the secrets required by the flow is considered sufficient to enable PDMv2. On the opposite, an unsuccessful decryption MUST result in dropping the PDMv2 DOH without further processing or, if configured to do so, might lead to throttling, filtering, and/or logging the activity of the other entity (Client or Server).¶
The present document specifies a methodology to enable this delegated trust, along with the Confidentiality and Integrity requirements, in the PDMv2 DOH.¶
We assume that PS and PC have verified the respective identities and the authorization to enable PDMv2 DOH on a set of devices under their responsibility: Secondary Servers (SS) and Secondary Clients (SC).¶
PS-PC¶
In other terms, if a PS would want to establish a pair with two PCs, it will have two different PMSs. PMS might be re-negotiated after a given amount of time [renegotiation TBD]¶
PS and PC exchange respectively the list of the SS and SC enabled to use PDMv2. The list can be:¶
Note:¶
PS-SS and PC-SC¶
Since there are multiple use-cases, we define 4 modes of operations:¶
The TrustedServers MoP has the benefit of requiring no additional steps to send and receive PDMv2 DOH, because each flow is protected by a SessionKey that can be derived autonomously by both the SC and the SS, without any interaction with the PS and PC, or any negotiation between the SS and the SC.¶
The possible vulnerabilities of the TrustedServers MoP are the following:¶
A possible mitigation is to split the Secondary Servers in different sub-groups. This is a scenario similar to the one of a PC negotiating PDMv2 access with different PSs.¶
The AsymmetricPoll MoP has the benefit of isolating each SS and each SC. Only the SS and SC involved in a communication can decrypt their flows.¶
The IBC MoP has the same security properties of the AsymmetricPoll MoP, and the advantage of not requiring any interaction between the Primary and the Secondary. The disadvantage is the requirement of performing a "pairing" session negotiation between the Secondaries.¶
It must be considered that, while secure, this MoP could be used to perform a resource exhaustion attack on the PairDeviceKey establishment. Hence, a device MUST NOT reply to an IP address that is not in the Secondary[client, server] list, and MUST NOT reply with negative acknowledgments (e.g., in case of an incorrect decoding).¶
PDMv2 greatly improves the privacy aspects of PDM by providing encryption.¶
The authors wish to thank NITK Surathkal for their support and assistance in coding and review. In particular Dr. Mohit Tahiliani and Abhishek Kumar (now with Google). Thanks also to Priyanka Sinha for her comments. Thanks to the India Internet Engineering Society (iiesoc.in), in particular Dhruv Dhody, for providing the funding for servers needed for protocol development.¶
Let's start with one client and one server.¶
+------------+ Derived Shared Secret +------------+
| Client | -----------------> | Server |
+------+-----+ +------+-----+
| |
V V
Client Secret Server Secret
¶
The Client and Server create public / private keys and derive a shared secret. Let's not consider Authentication or Certificates at this point.¶
What is stored at the Client and Server to be able to encrypt and decrypt packets? The shared secret or private key.¶
Since we only have one Server and one Client, then we don't need to have any kind of identifier for which private key to use for which Server or Client because there is only one of each.¶
Of course, this is a ludicrous scenario since no real organization of interest has only one server and one client.¶
So, let's try with multiple clients and one Primary server¶
+------------+
| Client 1 | --------+
+------------+ |
+------------+ +--->
| Client 2 | ----------> +------------+
+------+-----+ : | Server |
: : +------+-----+
: +---->
+------------+ |
| Client n | -------+
+------+-----+
¶
The Clients and Server create public / private keys and derive a shared secret. Each Client has a unique private key.¶
What is stored at the Client and Server to be able to encrypt and decrypt packets?¶
Clients each store a private key. Server stores: Client Identifier and Private Key.¶
Since we only have one Server and multiple Clients, then the Clients don't need to have any kind of identifier for which private key to use for which Server but the Server needs to know which private key to use for which Client. So, the Server has to store an identifier as well as the Key.¶
But, this also is a ludicrous scenario since no real organization of interest has only one server.¶
When we have multiple clients and multiple servers, then each not only does the Server need to know which key to use for which Client, but the Client needs to know which private key to use for which Server.¶
Based on this rationale, we have chosen a Primary Server / Primary Client topology.¶
In the Registration phase, the objective is to generate a shared secret that will be used in encryption and decryption during the Data Transfer phase. We have adopted a Primary-Secondary architecture to represent the clients and servers (see Section 4.1.1). The primary server and primary client perform Key Encapsulation Mechanism (KEM) [RFC9180] to generate a primary shared secret. The primary server shares this secret with secondary servers, whereas the primary client performs Key Derivation Function (KDF) [RFC9180] to share client- specific secrets to corresponding secondary clients. During the Data Transfer phase, the secondary servers generate the client-specific secrets on the arrival of the first packet from the secondary client.¶
The following steps describe the protocol flow:¶
Two commands are used between the primary client and the primary server to denote the setup and KEM phases. Along with this, we have a "req / resp" to indicate whether it's a request or response.¶
Between primary and secondary entities, we have one command to denote the sharing of the secret keys.¶
Note to RFC Editor: if this document does not obsolete an existing RFC, please remove this appendix before publication as an RFC.¶
Note to RFC Editor: please remove this appendix before publication as an RFC.¶