An RPKI and IPsec-based AS-to-AS Approach for Source Address Validation

Introduction Source address spoofing is the practice of using a source IP address without proper authorization from its owner. The basic internet routing architecture does not provide any defense against spoofing, so any system can send packets that claim any source address. This practice enables a variety of attacks, most notably volumetric DoS attacks as discussed in . There are many possible approaches to preventing address spoofing. describes three classes of Source Address Validation (SAV): Access Network, Intra-AS, and Inter-AS. Inter-AS SAV is the most challenging class, because different ASes have different policies and operate independently. Inter-AS SAV requires the different ASes to collaborate to verify the source address. However, in the absence of total trust between all ASes, Inter-AS SAV is a prerequisite to defeat source address spoofing. Despite years of effort, current Inter-AS SAV protocols are not widely deployed. An important reason is the difficulty of balancing the clear security benefits of partial implementations with the scalability of large-scale deployments. uRPF , for example, is a routing-based scheme that filters out spoofed traffic. In cases where the routing is dynamic or unknown, uRPF deployments must choose between false negatives (i.e. incomplete SAV) and false positives (i.e. broken routing). This document provides an RPKI- and IPsec-based approach to inter-AS source address validation (RISAV). RISAV is a cryptography-based SAV mechanism to reduce the spoofing of source addresses. In RISAV, the RPKI database acts as a root of trust for IPsec between participating ASes. Each pair of ASes uses IKEv2 to negotiate an IPsec Security Association (SA). Packets between those ASes are then protected by a modified IPsec Authentication Header (AH) or an Encapsulating Security Payload (ESP). IPsec authenticates the source address, allowing spoofed packets to be dropped at the border of the receiving AS.

Requirements Language 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 BCP14 when, and only when, they appear in all capitals, as shown here.

Terminology Commonly used terms in this document are described below.

ACS:: AS Contact Server, which is the logical representative of one AS and is responsible for delivering session keys and other information to ASBR.
Contact IP:: The IP address of the ACS.
ASBR:: AS border router, which is at the boundary of an AS.
SAV:: Source Address Validation, which verifies the source address of an IP packet and guarantee the source address is valid.

Overview The goal of this section is to provides the high level description of what RISAV is and how RISAV works.

What RISAV Is RISAV is a cryptographically-based inter-AS source address validation protocol that provides clear security benefits even at partial deployment. It aims to prove that each IP datagram was sent from inside the AS that owns its source address, defeating spoofing and replay attacks. It is light-weight and efficient, and provides incremental deployment incentives. At the source AS Border Router, RISAV adds a MAC to each packet that proves ownership of the packet's source address. At the recipient's ASBR, RISAV verifies and removes this MAC, recovering the unmodified original packet. The MAC is delivered in the Integrity Check Value (ICV) field of a modified IPsec AH, or as part of the normal IPsec ESP payload.

How RISAV Works RISAV uses IKEv2 to negotiate an IPsec security association (SA) between any two ASes. RPKI provides the binding relationship between AS numbers, IP ranges, contact IPs, and public keys. After negotiation, all packets between these ASes are secured by use of a modified AH header or a standard ESP payload. Before deploying RISAV, each AS sets a contact IP representative. When negotiating or consulting with one AS, the peer MUST first communicate with this contact IP. The AS MUST publish exactly one contact IP for each supported address family (i.e. IPv4 and/or IPv6) in the RPKI database. A typical workflow of RISAV is shown in .

RISAV workflow example.

RPKI process. The five Regional Internet Registries (RIR), authorized by IANA, use their root certificate to sign the Certificate Authority (CA) certificate of the Local Internet Registry (LIR), which is used to authorize the Autonomous System (AS) (sometimes indirectly via the Internet Service Provider (ISP)). When they obtain their own CA certificate, the AS would sign an End Entity (EE) certificate with a Route Origin Authorisation (ROA) which is a cryptographically signed object that states which AS are authorized to originate a certain prefix. This authenticated binding of the ASN to its IP prefixes is published in the RPKI database. This is a prerequisite for RISAV.
ACS EE certificate provisioning. The ACS would need its own EE certificate for IKEv2. This EE certificate is REQUIRED like the BGPsec Router Certificate defined in .
RISAV announcement. Each participating AS announces its support for RISAV in the RPKI database, including the IP address of its ACS (the "contact IP").
SA negotiation and delivery. The ACSes negotiate an SA using IKEv2. After synchronization, all ASBRs would get the SA, including the session key and other parameters.
IPsec communication. RISAV uses IPsec AH (i.e. "transport mode") for authentication of the IP source address by default. When an ASBR in AS A sends a packet to AS B, it uses the established IPsec channel to add the required AH header. The ASBR in AS B validates the AH header to ensure that the packet was not spoofed, and removes the header.

Control Plane The functions of the control plane of RISAV include:

Announcing that this AS supports RISAV.
Publishing contact IPs.
Performing IPsec session initialization (i.e. IKEv2).

These functions are achieved in two steps. First, each participating AS publishes a Signed Object in its RPKI Repository containing a RISAVAnnouncement: When a participating AS discovers another participating AS (via its regular sync of the RPKI database), it initiates an IKEv2 handshake between its own contact IP and the other AS's contact IP. This handshake MUST include an IKE_AUTH exchange that authenticates both ASes with their RPKI ROA certificates. Once this handshake is complete, each AS MUST activate RISAV on all outgoing packets, and SHOULD drop all non-RISAV traffic from the other AS after a reasonable grace period (e.g. 60 seconds). The "testing" field indicates whether this contact IP is potentially unreliable. When this field is set to true, other ASes MUST fall back to ordinary operation if IKE negotiation fails. Otherwise, the contact IP is presumed to be fully reliable, and other ASes SHOULD drop all non-RISAV traffic from this AS if IKE negotiation fails (see ). For more information about RPKI, see .

Disabling RISAV To disable RISAV, a participating AS MUST perform the following steps in order:

Stop requiring RISAV authentication of incoming packets.
Remove the RISAVAnnouncement from the RPKI Repository.
Wait at least 24 hours.
Stop sending RISAV and shut down the contact IP.

Conversely, if any AS no longer publishes a RISAVAnnouncement, other ASes MUST immediately stop sending RISAV to that AS, but MUST NOT delete any negotiated Tunnel Mode SAs for at least 24 hours, in order to continue to process encrypted incoming traffic.

TODO: Discuss changes to the contact IP, check if there are any race conditions between activation and deactivation, IKEv2 handshakes in progress, SA expiration, etc.

SA has its own expiration time and IKE has its keepalive mechanism. In abnormal case, i.e. the connection is failed after the IKE handshake is established, SA will be always in effect during its lifetime until it expires or the IKE keepalive is failed. In normal case, i.e. the connection is actively down, SA will be expired and RISAV will be disabled immediately.

OPEN QUESTION: Does IKEv2 have an authenticated permanent rejection option that would help here?

Data Plane All the ASBRs of the AS are REQUIRED to enable RISAV. The destination ASBR uses the IPsec SPI to locate the correct SA. As defined in , the Security Association Database (SAD) stores all the SAs. Each data item in the SAD includes a cryptographic algorithm (e.g. HMAC-SHA-256), its corresponding key, and other relevant parameters. When an outgoing packet arrives at the source ASBR, its treatment depends on the source and destination address. If the source address belongs to the AS in which the ASBR is located, and the destination address is in an AS for which the ASBR has an active RISAV SA, then the packet needs to be modified for RISAV. The modification that is applied depends on whether IPsec "transport mode" or "tunnel mode" is active. This is determined by the presence or absence of the USE_TRANSPORT_MODE notification in the IKEv2 handshake. RISAV implementations MUST support transport mode, and MAY support tunnel mode.

OPEN QUESTION: How do peers express a preference or requirement for transport or tunnel mode?

When a packet arrives at the destination ASBR, it will check the destination address and the source address. If the destination belongs to the AS in which the destination ASBR is located, and the source address is in an AS with which this AS has an active RISAV SA, then the packet is subject to RISAV processing. To avoid DoS attacks, participating ASes MUST drop any outgoing packet to the contact IP of another AS. Only the AS operator's systems (i.e. the ACS and ASBRs) are permitted to send packets to the contact IPs of other ASes. ASBRs MAY drop inbound packets to the contact IP from non-participating ASes.

Transport Mode To avoid conflict with other uses of IPsec (), RISAV defines its own variant of the IPsec Authentication Header (AH). The RISAV-AH header format is shown in .

RISAV-AH Format. This format is identical to IPsec standard AH except that the Sequence Number is omitted, because RISAV is presumed to be a "multi-sender SA" for which anti-replay defense is not supported (). This change saves 8 octets when the ICV is 16, 24, or 32 octets. For a 16-octet ICV (most common), RISAV-AH adds 24 octets to each packet. The RISAV-AH header is only for AS-to-AS communication. ASes MUST strip off all RISAV-AH headers for packets whose destination is inside the AS, even if the AS is not currently inspecting the ICV values. In transport mode, each AS's SA Database (SAD) is indexed by SPI and counterpart AS, regardless of the source and destination IPs.

Tunnel Mode In tunnel mode, a RISAV sender ASBR wraps each outgoing packet in an ESP payload. Each ASBR uses its own source address, and sets the destination address to the contact IP of the destination AS. The contact IP decrypts all IPsec traffic to recover the original packets, which are forwarded to the correct destination. After decryption, the receiving AS MUST check that the source IP and destination IP are in the same AS as the outer source and destination, respectively. In tunnel mode, each ASBR maintains its own copy of the SA Database (SAD). Each copy of the SAD is indexed by SPI and counterpart AS. If a valid ESP packet is received from an unknown IP address, the receiving AS SHOULD allocate a new replay defense window, subject to resource constraints. This allows replay defense to work as usual. (If the contact IP is implemented as an ECMP cluster, effective replay defense may require consistent hashing.) Tunnel mode imposes a space overhead of 73 octets in IPv6.

PROBLEM: ESP doesn't protect the source IP, so a packet could be replayed by changing the source IP. Can we negotiate an extension to ESP that covers the IP header? Or could we always send from the contact IP and encode the ASBR ID in the low bits of the SPI?

Possible Extensions This section presents potential additions to the design.

TODO: Remove this section once we have consensus on whether these extensions are worthwhile.

Header-only authentication RISAV-AH, like standard IPsec AH, authenticates the whole constant part of a packet, including the entire payload. To improve efficiency, we could define an IKE parameter to negotiate a header-only variant of transport mode that only authenticates the IP source address, IP destination address, etc. This would likely result in a 10-30x decrease in cryptographic cost compared to standard IPsec. However, it would also offer no SAV defense against any attacker who can view legitimate traffic. An attacker who can read a single authenticated packet could simply replace the payload, allowing it to issue an unlimited number of spoofed packets.

Time-based key rotation Each IKEv2 handshake negotiates a fixed shared secret, known to both parties. In some cases, it might be desirable to rotate the shared secret frequently:

In transport mode, frequent rotation would limit how long a single packet can be replayed by a spoofing attacker.
If the ASBRs are less secure than the ACS, frequent rotation could limit the impact of a compromised ASBR.

However, increasing the frequency of IKEv2 handshakes would increase the burden on the ACS. One alternative possibility is to use a state machine. The state machine runs and triggers the state transition when time is up. The tag is generated in the process of state transition as the side product. The two ACS in peer AS respectively before data transmission will maintain one state machine pair for each bound. The state machine runs simultaneously after the initial state, state transition algorithm, and state transition interval are negotiated, thus they generate the same tag at the same time. Time triggers state transition which means the ACS MUST synchronize the time to the same time base using like NTP defined in . For the tag generation method, it MUST be to specify the initial state and initial state length of the state machine, the identifier of a state machine, state transition interval, length of generated Tag, and Tag. For the SA, they will transfer all these payloads in a secure channel between ACS and ASBRs, for instance, in ESP . It is RECOMMENDED to transfer the tags rather than the SA for security and efficiency considerations. The initial state and its length can be specified at the Key Exchange Payload with nothing to be changed. The state machine identifier is the SPI value as the SPI value is uniquely in RISAV. The state transition interval and length of generated Tag should be negotiated by the pair ACS, which will need to allocate one SA attribute. The generated Tag will be sent from ACS to ASBR in a secure channel which MAY be, for example, ESP .

Static Negotiation The use of IKEv2 between ASes might be fragile, and creates a number of potential race conditions (e.g. if the RPKI database contents change during the handshake). It is also potentially costly to implement, requiring O(N^2) network activity for N participating ASes. If these challenges prove significant, one alternative would be to perform the handshake statically via the RPKI database. For example, static-static ECDH would allow ASes to agree on shared secrets simply by syncing the RPKI database. Static negotiation makes endpoints nearly stateless, which simplifies the provisioning of ASBRs. However, it requires inventing a novel IPsec negotiation system, so it seems best to try a design using IKEv2 first.

Security Consideration

Threat models In general, RISAV seeks to provide a strong defense against arbitrary active attackers who are external to the source and destination ASes. However, different RISAV modes and configurations offer different security properties.

Replay attacks In Transport Mode, off-path attackers cannot spoof the source IPs of a participating AS, but any attacker with access to valid traffic can replay it (from anywhere), potentially enabling DoS attacks by replaying expensive traffic (e.g. TCP SYNs, QUIC Initials). ASes that wish to have replay defense, and are willing to pay the extra data-plane costs, should prefer tunnel mode.

Downgrade attacks An on-path attacker between two participating ASes could attempt to defeat RISAV by blocking IKEv2 handshakes to the Contact IP of a target AS. If the AS initiating the handshake falls back to non-RISAV behavior after a handshake failure, this enables the attacker to remove all RISAV protection. This vulnerable behavior is required when the "testing" flag is set, but is otherwise discouraged.

Incremental benefit from partial deployment RISAV provides significant security benefits even if it is only deployed by a fraction of all ASes. This is particularly clear in the context of reflection attacks. If two networks implement RISAV, no one in any other network can trigger a reflection attack between these two networks. Thus, if X% of ASes (selected at random) implement RISAV, participating ASes should see an X% reduction in reflection attack traffic volume.

Compatibility

With end-to-end IPsec When RISAV is used in transport mode, there is a risk of confusion between the RISAV AH header and end-to-end AH headers used by applications. This risk is particularly clear during transition periods, when the recipient is not sure whether the sender is using RISAV or not. To avoid any such confusion, RISAV's transport mode uses a specialized RISAV-AH header. (In tunnel mode, no such confusion is possible.)

With other SAV mechanisms RISAV is independent from intra-domain SAV and access-layer SAV, such as or SAVI . When these techniques are used together, intra-domain and access-layer SAV checks MUST be enforced before applying RISAV.

Operational Considerations

Reliability The ACS, represented by a contact IP, must be a high-availability, high-performance service to avoid outages. This might be achieved by electing one distinguished ASBR as the ACS. The distinguished ASBR acting as an ACS will represent the whole AS to communicate with peer AS's ACS. This election takes place prior to the IKE negotiation. In this arrangement, an ASBR MUST be a BGP speaker before it is elected as the distinguished ASBR.

Synchronizing Multiple ASBRs In RISAV, all ASBRs of each AS must have the same Security Associations, because the recipient does not keep distinct state for each sending ASBR (except for the replay window in tunnel mode). For example, ASBRs cannot perform IKE negotiation independently. Instead, the ACS is the entity that represents the AS to negotiate associations with other ASes. To ensure coherent behavior across the AS, the ACS MUST deliver each SA to all ASBRs in the AS immediately after it is negotiated. RISAV does not standardize a mechanism for this update broadcast. During the SA broadcast, ASBRs will briefly be out of sync. RISAV recommends a grace period to prevent outages during the update process.

Performance RISAV requires participating ASes to perform symmetric cryptography on every RISAV-protected packet that they originate or terminate. This will require significant additional compute capacity that may not be present on existing networks. However, until most ASes actually implement RISAV, the implementation cost for the few that do is greatly reduced. For example, if 5% of networks implement RISAV, then participating networks will only need to apply RISAV to 5% of their traffic. Thanks to broad interest in optimization of IPsec, very high performance implementations are already available. For example, as of 2021 an IPsec throughput of 1 Terabit per second was achievable using optimized software on a single server .

MTU

TODO: Figure out what to say about MTU, PMTUD, etc. Perhaps an MTU probe is required after setup? Or on an ongoing basis?

NAT scenario As all the outer IP header should be the unicast IP address, NAT-traversal mode is not necessary in inter-AS SAV.

IANA Consideration IF APPROVED IANA is requested to add the following entry to the Assigned Internet Protocol Numbers registry:

Decimal: $TBD
Keyword: RISAV-AH
Protocol: AS-to-AS Authentication Header
IPv6 Extension Header: Y
Refrence: (This document)