Internet-Draft | Segment Routing IPv6 Security Considerat | August 2024 |
Buraglio, et al. | Expires 22 February 2025 | [Page] |
SRv6 is a traffic engineering, encapsulation and steering mechanism utilizing IPv6 addresses to identify segments in a pre-defined policy. This document discusses security considerations in SRv6 networks, including the potential threats and the possible mitigation methods. The document does not define any new security protocols or extensions to existing protocols.¶
This note is to be removed before publishing as an RFC.¶
The latest revision of this draft can be found at https://buraglio.github.io/draft-bdmgct-spring-srv6-security/draft-bdmgct-spring-srv6-security.html. Status information for this document may be found at https://datatracker.ietf.org/doc/draft-ietf-spring-srv6-security/.¶
Discussion of this document takes place on the Source Packet Routing in Networking Working Group mailing list (mailto:spring@ietf.org), which is archived at https://mailarchive.ietf.org/arch/browse/spring/. Subscribe at https://www.ietf.org/mailman/listinfo/spring/.¶
Source for this draft and an issue tracker can be found at https://github.com/buraglio/draft-bdmgct-spring-srv6-security.¶
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 22 February 2025.¶
Copyright (c) 2024 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.¶
Segment Routing (SR) [RFC8402] utilizing an IPv6 data plane is a source routing model that leverages an IPv6 underlay and an IPv6 extension header called the Segment Routing Header (SRH) [RFC8754] to signal and control the forwarding and path of packets by imposing an ordered list of path details that are processed at each hop along the signaled path. Because SRv6 is fundamentally bound to the IPv6 protocol, and because of the reliance on a new header there are security considerations which must be noted or addressed in order to operate an SRv6 network in a reliable and secure manner. Specifically, some primary properties of SRv6 that affect the security considerations are:¶
SRv6 may use the SRH which is a type of Routing Extension Header defined by [RFC8754]. Some security considerations of the SRH are discussed in [RFC5095] section 5 and [RFC8754] section 7.¶
SRv6 uses the IPv6 data-plane, and therefore known security considerations of IPv6 [RFC9099] are applicable to SRv6 as well.¶
While SRv6 uses what appear to be typical IPv6 addresses, the address space is processed differently by segment endpoints. A typical IPv6 unicast address is composed of a network prefix, host identifier, and a subnet mask. A typical SRv6 segment identifier (SID) consists of a locator, a function identifier, and optionally, function arguments (LOC:FUNCT:ARG [RFC8986]). The locator must be routable, which enables both SRv6 capable and incapable devices to participate in forwarding, either as normal IPv6 unicast or SRv6. The capability to operate in environments that may have gaps in SRv6 support allows the bridging of islands of SRv6 devices with standard IPv6 unicast routing.¶
This document describes various threats to SRv6 networks and also presents existing approaches to avoid or mitigate the threats.¶
The following IETF RFCs were selected for security assessment as part of this effort:¶
[RFC8986] : "Segment Routing over IPv6 (SRv6) Network Programming"¶
[RFC9491] : "Integration of the Network Service Header (NSH) and Segment Routing for Service Function Chaining (SFC)"¶
[RFC9524] : "Segment Routing Replication for Multipoint Service Delivery"¶
We note that SRv6 is under active development and, as such, the above documents might not cover all protocols employed in an SRv6 deployment.¶
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.¶
This section introduces the threat model that is used in this document. The model is based on terminology from the Internet threat model [RFC3552], as well as some concepts from [RFC9055], [RFC7384] and [RFC9416]. Details regarding inter-domain segment routing (SR) are out of scope for this document.¶
An internal attacker in the context of SRv6 is an attacker who is located within an SR domain. Specifically, an internal attacker either has access to a node in the SR domain, or is located on an internal path between two nodes in the SR domain. External attackers, on the other hand, are not within the SR domain.¶
On-path attackers are located in a position that allows interception, modification or dropping of in-flight packets, as well as insertion (generation) of packets. Off-path attackers can only attack by insertion of packets.¶
The following figure depicts the attacker types according to the taxonomy above. As illustrated in the figure, on-path attackers are located along the path of the traffic that is under attack, and therefore can listen, insert, delete, modify or replay packets in transit. Off-path attackers can insert packets, and in some cases can passively listen to some traffic, such as multicast transmissions.¶
on-path on-path off-path off-path external internal internal external attacker attacker attacker attacker | | |__ | | SR __ | __ _/| \ | | domain / \_/ | \_/ v \__ v | \ | X \ X v / v \ ----->X---------->O------>X------->O------->O----> ^\ ^ /^ | \___/\_ /\_ | _/\__/ | | \__/ | | | | | SR SR SR ingress endpoint egress node node
In the current threat model the SR domain defines the boundary that distinguishes internal from external threats. As specified in [RFC8402]:¶
By default, SR operates within a trusted domain. Traffic MUST be filtered at the domain boundaries. The use of best practices to reduce the risk of tampering within the trusted domain is important. Such practices are discussed in [RFC4381] and are applicable to both SR-MPLS and SRv6.¶
In the context of the current document it is assumed that SRv6 is deployed within a limited domain [RFC8799] with filtering at the domain boundaries, forming a trusted domain with respect to SRv6. Thus, external attackers are outside the trusted domain. Specifically, an attack on one domain that is invoked from within a different domain is considered an external attack in the context of the current document.¶
Following the spirit of [RFC8402], the current document mandates a filtering mechanism that eliminates the threats from external attackers. This approach limits the scope of the attacks described in this document to within the domain (i.e., internal attackers).¶
It should be noted that in some threat models the distinction between internal and external attackers depends on whether an attacker has access to a cryptographically secured (encrypted or authenticated) domain. Specifically, in some of these models there is a distinction between an attacker who becomes internal by having physical access, for example by plugging into an active port of a network device, and an attacker who has full access to a legitimate network node, including for example encryption keys if the network is encrypted. The current model does not distinguish between these two types of attackers and there is no assumption about whether the SR domain is cryptographically secured or not. Thus, some of the attacks that are described in the next section can be mitigated by cryptographic means, as further discussed in Section 7.3.¶
One of the important aspects of a threat analysis is the potential impact of each threat. SRv6 allows for the sending of IPv6 packets via arbitrary paths. An attack on SRv6 may cause packets to traverse arbitrary paths within an SR domain. This may allow an attacker to perform a number of attacks on the victim networks and hosts that would be mostly unfeasible for a non-SRv6 environment.¶
The threat model in [ANSI-Sec] classifies threats according to their potential impact, defining six categories. For each of these categories we briefly discuss its applicability to SRv6 attacks.¶
Unauthorized Access: an attack that results in unauthorized access might be achieved by having an attacker leverage SRv6 to circumvent security controls as a result of security devices being unable to enforce security policies in the presence of IPv6 Extension Headers (see [RFC9098]), or by directing packets through paths where packet-filtering policies are not enforced.¶
Masquerade: various attacks that result in spoofing or masquerading are possible in IPv6 networks (e.g., [RFC9098]). However, these attacks are not specific to SRv6, and are therefore not within the scope of this document.¶
System Integrity: attacks on SRv6 can manipulate the path and the processing that the packet is subject to, thus compromising the integrity of the system. Furthermore, an attack that compromises the control plane and/or the management plane is also a means of impacting the system integrity.¶
Communication Integrity: SRv6 attacks may cause packets to be forwarded through paths that the attacker controls, which may facilitate other attacks that compromise the integrity of user data. Integrity protection of user data, which is implemented in higher layers, avoids these aspects, and therefore communication integrity is not within the scope of this document.¶
Confidentiality: as in communication integrity, packets forwarded through unintended paths may traverse nodes controlled by the attacker. Since eavesdropping to user data can be avoided by using encryption in higher layers, it is not within the scope of this document. However, eavesdropping to a network that uses SRv6 allows the attacker to collect information about SR endpoint addresses, SR policies, and network topologies, is a specific form of reconnaissance¶
Denial of Service: the availability aspects of SRv6 include the ability of attackers to leverage SRv6 as a means for compromising the performance of a network or for causing Denial of Service (DoS). Compromising the availability of the system can be achieved by sending multiple SRv6-enabled packets to/through victim nodes, where the SRv6-enabled packets result in a negative performance impact of the victim systems (see [RFC9098] for further details). Alternatively, an attacker might achieve attack amplification by causing packets to "bounce" multiple times between a set of victim nodes, with the goal of exhausting processing resources and/or bandwidth (see [CanSecWest2007] for a discussion of this type of attack).¶
Section 6 discusses specific implementations of these attacks, and possible mitigations are discussed in Section 7.¶
Packet manipulation and processing attacks can be implemented by performing a set of one or more basic operations. These basic operations (abstractions) are as follows:¶
Passive listening: an attacker who reads packets off the network can collect information about SR endpoint addresses, SR policies and the network topology. This information can then be used to deploy other types of attacks.¶
Packet replaying: in a replay attack the attacker records one or more packets and transmits them at a later point in time.¶
Packet insertion: an attacker generates and injects a packet to the network. The generated packet may be maliciously crafted to include false information, including for example false addresses and SRv6-related information.¶
Packet deletion: by intercepting and removing packets from the network, an attacker prevents these packets from reaching their destination. Selective removal of packets may, in some cases, cause more severe damage than random packet loss.¶
Packet modification: the attacker modifies packets during transit.¶
This section describes attacks that are based on packet manipulation and processing, as well as attacks performed by other means.¶
An attacker can modify a packet while it is in transit in a way that directly affects the packet's SR policy. The modification can affect the destination address of the IPv6 header and/or the SRH. In this context SRH modification may refer to inserting an SRH, removing an SRH, or modifying some fields of an existing SRH.¶
Modification of an existing SRH can be further classified into several possible attacks. Specifically, the attack can include adding one or more SIDs to the segment list, removing one or more SIDs or replacing some SIDs with different SIDs. Another possible type of modification is by adding, removing or modifying TLV fields in the SRH.¶
When an SRH is present modifying the destination address (DA) of the IPv6 header affects the active segment. However, DA modification can affect the SR policy even in the absence of an SRH. One example is modifying a DA which is used as a Binding SID [RFC8402]. Another example is modifying a DA which represents a compressed segment list [I-D.ietf-spring-srv6-srh-compression]. SRH compression allows encoding multiple compressed SIDs within a single 128-bit SID, and thus modifying the DA can affect one or more hops in the SR policy.¶
An SR modification attack can be performed by on-path attackers. As discussed in Section 4, it assumed that filtering is deployed at the domain boundaries, thus limiting the ability of implementing SR modification attacks to on-path internal attackers.¶
The SR modification attack allows an attacker to change the SR policy that the packet is steered through and thus to manipulate the path and the processing that the packet is subject to.¶
Specifically, the SR modification attack can impact the network and the forwarding behavior of packets in one or more of the following ways:¶
An attacker can manipulate the DA and/or SRH in order to avoid a specific node or path. This approach can be used, for example, for bypassing the billing service or avoiding access controls and security filters.¶
The packet can be manipulated to avert packets to a specific path. This attack can result in allowing various unauthorized services such as traffic acceleration. Alternatively, an attacker can avert traffic to be forwarded through a specific node that the attacker has access to, thus facilitating more complex on-path attacks such as passive listening, recon and various man-in-the-middle attacks. It is noted that the SR modification attack is performed by an on-path attacker who has access to packets in transit, and thus can implement these attacks directly. However, SR modification is relatively easy to implement and requires low processing resources by an attacker, while it facilitates more complex on-path attacks by averting the traffic to another node that the attacker has access to and has more processing resources.¶
SR modification may cause a packet to be forwarded to a point in the network where it can no longer be forwarded, causing the packet to be discarded.¶
An attacker can trigger a specific endpoint behavior by modifying the destination address and/or SIDs in the segment list. This attack can be invoked in order to manipulate the path or in order to exhaust the resources of the SR endpoint.¶
An attacker can add SIDs to the segment list in order to increase the number hops that each packet is forwarded through and thus increase the load on the network. For example, a set of SIDs can be inserted in a way that creates a forwarding loop ([RFC8402], [RFC5095]) and thus loads the nodes along the loop. Network programming can be used in some cases to manipulate segment endpoints to perform unnecessary functions that consume processing resources. Path inflation, malicious looping and unnecessary instructions have a common outcome, resource exhaustion, which may in severe cases cause Denial of Service (DoS).¶
An on-path attacker can passively listen to packets and specifically to the SRv6-related information that is conveyed in the IPv6 header and the SRH. This approach can be used for collecting information about SIDs and policies, and thus to facilitate mapping the structure of the network and its potential vulnerabilities.¶
A recon attack is limited to on-path internal attackers.¶
It is assumed that the SRv6 domain is filtered in a way that prevents any leaks of explicit SRv6 routing information through the boundaries of the administrative domain. External attackers can only collect SRv6-related data in a malfunctioning network in which SRv6-related information is leaked through the boundaries of an SR domain.¶
In this attack packets are inserted (injected) into the network with a segment list that defines a specific SR policy. The attack can be applied either by using synthetic packets or by replaying previously recorded packets.¶
Packet insertion can be performed by internal attackers, either on-path or off-path. In the case of a replay attack, recording packets in-flight requires on-path access and the recorded packets can later be injected either from an on-path or an off-path location.¶
SRv6 domains are assumed to be filtered in a way that mitigates insertion attacks from external attackers.¶
The main impact of this attack is resource exhaustion which compromises the availability of the network, as described in Section 6.2.3.¶
Depending on the control plane protocols used in a network, it is possible to use the control plane as a way of compromising the network. For example, an attacker can advertise SIDs in order to manipulate the SR policies used in the network. A wide range of attacks can be implemented, including injecting control plane messages, selectively removing legitimate messages, replaying them or passively listening to them.¶
A compromised management plane can also facilitate a wide range of attacks, including manipulating the SR policies or compromising the network availability.¶
Control plane attacks can be performed by internal attackers. Injection can be performed by off-path attackers, while removal, replaying and listening require on-path access. The scope of management attacks depends on the specific management protocol and architecture.¶
It is assumed that SRv6 domain boundary filtering is used for mitigating potential control plane and management plane attacks from external attackers. Segment routing does not define any specific security mechanisms in existing control plane or management plane protocols. However, existing control plane and management plane protocols use authentication and security mechanisms to validate the authenticity of information.¶
A compromised control plane or management plane can impact the network in various possible ways. SR policies can be manipulated by the attacker to avoid specific paths or to prefer specific paths, as described in Section 6.2.3. Alternatively, the attacker can compromise the availability, either by defining SR policies that load the network resources, as described in Section 6.2.3, or by blackholing some or all of the SR policies. A passive attacker can use the control plane or management plane messages as a means for recon, similarly to Section 6.2.3.¶
Various attacks which are not specific to SRv6 can be used to compromise networks that deploy SRv6. For example, spoofing is not specific to SRv6, but can be used in a network that uses SRv6. Such attacks are outside the scope of this document.¶
Because SRv6 is completely reliant on IPv6 for addressing, forwarding, and fundamental networking basics, it is potentially subject to any existing or emerging IPv6 vulnerabilities [RFC9099], however, this is out of scope for this document.¶
This section presents methods that can be used to mitigate the threats and issues that were presented in previous sections. This section does not introduce new security solutions or protocols.¶
SRv6 packets rely on the routing header in order to steer traffic that adheres to a defined SRv6 traffic policy. Thus, SRH filtering can be enforced at the ingress and egress nodes of the SR domain, so that packets with an SRH cannot be forwarded into the SR domain or out of the SR domain. Specifically, such filtering is performed by detecting Next Header 43 (Routing Header) with Routing Type 4 (SRH).¶
Because of the methodologies used in SID compression [I-D.ietf-spring-srv6-srh-compression], SRH compression does not necessarily use an SRH. In practice this means that when compressed segment lists are used without an SRH, filtering based on the Next Header is not relevant, and thus filtering can only be applied based on the address range, as described below.¶
The IPv6 destination address can be filtered at the SR ingress node in order to mitigate external attacks. An ingress packet with a destination address that defines an active segment with an SR endpoint in the SR domain is filtered.¶
In order to apply such a filtering mechanism the SR domain needs to have an allocated address range that can be detected and enforced by the SR ingress, for example by using ULA addresses, or preferably, by using the IANA special use prefix [IANAIPv6SPAR] for SRv6, 5f00::/16.¶
Note that the use of GUA addressing in data plane programming could result in an fail open scenario when appropriate border filtering is not implemented or supported.¶
Packets steered in an SR domain are often encapsulated in an IPv6 encapsulation. This mechanism allows for encapsulation of both IPv4 and IPv6 packets. Encapsulation of packets at the SR ingress node and decapsulation at the SR egress node mitigates the ability of external attackers to impact SR steering within the domain.¶
The SRH can be secured by an HMAC TLV, as defined in [RFC8754]. The HMAC is an optional TLV that secures the segment list, the SRH flags, the SRH Last Entry field and the IPv6 source address. A pre-shared key is used in the generation and verification of the HMAC.¶
Using an HMAC in an SR domain can mitigate some of the SR Modification Attacks (Section 6.2). For example, the segment list is protected by the HMAC.¶
The following aspects of the HAMC should be considered:¶
The HMAC TLV is OPTIONAL.¶
While it is presumed that unique keys will be employed by each participating node, in scenarios where the network resorts to manual configuration of pre-shared keys, the same key might be reused by multiple systems as an (incorrect) shortcut to keeping the problem of pre-shared key configuration manageable.¶
An internal attacker who does not have access to the pre-shared key can capture legitimate packets, and later replay the SRH and HMAC from these recorded packets. This allows the attacker to insert the previously recorded SRH and HMAC into a newly injected packet. An on-path internal attacker can also replace the SRH of an in-transit packet with a different SRH that was previously captured.¶
[RFC9288] provides recommendations on the filtering of IPv6 packets containing IPv6 extension headers at transit routers. SRv6 relies on the routing header (RH4). Because the technology is reasonably new, many platforms, routing and otherwise, do not possess the capability to filter and in some cases even provide logging for IPv6 next-header 43 Routing type 4.¶
When an SRv6 packet is forwarded in the SRv6 domain, its destination address changes constantly, the real destination address is hidden. Security devices on SRv6 network may not learn the real destination address and fail to take access control on some SRv6 traffic.¶
The security devices on SRv6 networks need to take care of SRv6 packets. However, the SRv6 packets usually use loopback address of the PE device as a source address. As a result, the address information of SR packets may be asymmetric, resulting in improper filter traffic problems, which affects the effectiveness of security devices. For example, along the forwarding path in SRv6 network, the SR-aware firewall will check the association relationships of the bidirectional VPN traffic packets. And it is able to retrieve the final destination of SRv6 packet from the last entry in the SRH. When the <source, destination> tuple of the packet from PE1 to PE2 is <PE1-IP-ADDR, PE2-VPN-SID>, and the other direction is <PE2-IP-ADDR, PE1-VPN-SID>, the source address and destination address of the forward and backward VPN traffic are regarded as different flow. Eventually, the legal traffic may be blocked by the firewall.¶
SRv6 is commonly used as a tunneling technology in operator networks. To provide VPN service in an SRv6 network, the ingress PE encapsulates the payload with an outer IPv6 header with the SRH carrying the SR Policy segment List along with the VPN service SID. The user traffic towards SRv6 provider backbone will be encapsulated in SRv6 tunnel. When constructing an SRv6 packet, the destination address field of the SRv6 packet changes constantly and the source address field of the SRv6 packet is usually assigned using an address on the originating device, which may be a host or a network element depending on configuration. This may affect the security equipment and middle boxes in the traffic path. Because of the existence of the SRH, and the additional headers, security appliances, monitoring systems, and middle boxes could react in different ways if do not incorporate support for the supporting SRv6 mechanisms, such as the IPv6 Segment Routing Header (SRH) [RFC8754]. Additionally, implementation limitations in the processing of IPv6 packets with extension headers may result in SRv6 packets being dropped [RFC7872],[RFC9098].¶
The security considerations of SRv6 are presented throughout this document.¶
This document has no IANA actions.¶
This section lists topics that will be discussed further before deciding whether they need to be included in this document, as well as some placeholders for items that need further work.¶
The following references may be used in the future: RFC9256 [RFC8986]¶
SRH compression¶
Spoofing¶
Path enumeration¶
Infrastructure and topology exposure: this seems like a non-issue from a WAN perspective. Needs more thought - could be problematic in a host to host scenario involving a WAN and/or a data center fabric.¶
Terms that may be used in a future version: Locator Block, FRR, uSID¶
L4 checksum: [RFC8200] specifies that when the Routing header is present the L4 checksum is computed by the originating node based on the IPv6 address of the last element of the Routing header. When compressed segment lists [I-D.ietf-spring-srv6-srh-compression] are used, the last element of the Routing header may be different than the Destination Address as received by the final destination. Furthermore, compressed segment lists can be used in the Destination Address without the presence of a Routing header, and in this case the IPv6 Destination address can be modified along the path. As a result, some existing middleboxes which verify the L4 checksum might miscalculate the checksum. This issue is currently under discussion in the SPRING WG.¶
Segment Routing Header figure: the SRv6 Segment Routing Header (SRH) is defined in [RFC8754].¶
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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Next Header | Hdr Ext Len | Routing Type | Segments Left | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Last Entry | Flags | Tag | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Segment List[0] (128 bits IPv6 address) | | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | | ... | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Segment List[n] (128 bits IPv6 address) | | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
The authors would like to acknowledge the contributions from Andrew Alston, Dale Carder, Bruno Decraene, Dhruv Dhody, Joel Halpern, Bruno Hassanov, Alvaro Retana, Eric Vyncke, and Zafar Ali.¶