Source Address Validation in Intra-domain Networks Gap Analysis, Problem Statement, and Requirements

Introduction Source Address Validation (SAV) is important for defending against source address spoofing attacks. Network operators can implement SAV mechanisms at multiple levels: access-network SAV, intra-domain SAV, and inter-domain SAV (see ). Access-network SAV (e.g., SAVI [RFC7513], IP Source Guard (IPSG) based on DHCP snooping , and Cable Source-Verify ) is typically deployed on switches inside the access network to prevent a host from using the source address of another host in the same network segment. When access-network SAV is not universally deployed, intra-domain SAV on routers of the Autonomous System (AS) can increase the defense in depth by blocking spoofing traffic as close to the source as possible. This document focuses only on the analysis of intra-domain SAV (also known as intra-AS SAV). Unlike inter-domain SAV (also known as inter-AS SAV) which requires information (e.g., Border Gateway Protocol (BGP) data) provided by other ASes to determine SAV rules, intra-domain SAV for an AS determines SAV rules solely by the AS itself with local information (e.g., local configuration or Interior Gateway Protocol (IGP) data). Specifically, as illustrated in , intra-domain SAV for an AS achieves two goals: i) prevent a sub network from using a source address out of the sub network or prevent a directly-connected host from using a source address out of the network segment; and ii) prevent spoofed data packets coming from the rest of the Internet that use a source address of the local AS. A sub network is part of the AS. It consists of routers and hosts and may run a routing protocol among its routers. It only originates traffic and is connected to one or more routers of the AS for Internet connectivity.

Two Goals of intra-domain SAV | Router || |+----------------------------+ +--------+| +----------------------------------------------------------------+ Case ii: The AS receives spoofed data packets using a source address of the local AS from the rest of the Internet Goal ii: If the AS deploys intra-domain SAV, the spoofed data packets can be blocked Spoofed data packets using a source address +----+ of the local AS +--------------------------+ | AS |<---------------------| The rest of the Internet | +----+ +--------------------------+ ]]> In the following, this document provides a gap analysis of the current operational intra-domain SAV mechanisms, concludes key problems to solve, and proposes basic requirements for future ones.

Terminology Sub Network: A sub network is part of the AS. It consists of routers and hosts and may run the routing protocol among its routers. It only originates traffic and is connected to one or more routers of the AS for Internet connectivity. SAV Rule: The rule in a router that describes the mapping relationship between a source address (prefix) and the valid incoming interface(s). It is used by a router to make SAV decisions. Improper Block: The validation results that the packets with legitimate source addresses are blocked improperly due to inaccurate SAV rules. Improper Permit: The validation results that the packets with spoofed source addresses are permitted improperly due to inaccurate SAV rules. SAV-specific Information: The information specialized for SAV rule generation.

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 BCP 14 when, and only when, they appear in all capitals, as shown here.

Current Operational Intra-domain SAV Mechanisms Although BCP38 and BCP84 provides multiple ingress filtering methods which are typically used for inter-domain SAV, some of them have also been used for intra-domain SAV in practice. This section introduces the current operational mechanisms used to implement intra-domain SAV.

Access Control List (ACL) is a SAV filter that checks the source address of every data packet against a list of acceptable or unacceptable prefixes. By performing the ACL rule at a router interface, every data packet received at this interface that does not match the ACL rule will be blocked. It can be used at router interfaces facing a sub network or a set of hosts, only allowing data packets using an acceptable source address. It is also usually used on router interfaces facing the rest of the Internet, blocking data packets using an unacceptable source address, such as internal source addresses owned by the local AS . The ACL rule is typically configured and updated manually, so it is critical to update ACL rules in time when the set of prefixes changes.
Strict uRPF implements a SAV filter in an automatic way by checking the source address of every data packet against the router's local Forwarding Information Base (FIB). The router deploying strict uRPF accepts a data packet only when i) the local FIB contains a prefix covering the packet's source address and ii) the corresponding outgoing interface for the prefix in the FIB matches the packet's incoming interface. Otherwise, the packet will be blocked. Strict uRPF is often used to block spoofed data packets originated from a sub network or a directly-connected host.
Loose uRPF also uses the local FIB to implement SAV but checks only for the existence of the prefix. Loose uRPF accepts a data packet if the router's local FIB contains a prefix covering the packet's source address regardless of the interface from which the packet is received. Routers can use loose uRPF to block spoofed data packets using a non-global or non-routed source address .

EFP-uRPF is another advanced SAV mechanism but it is specifically designed for inter-domain SAV. It implements a SAV filter on router interfaces facing a customer AS by using BGP data provided by other ASes. This document does not analyze EFP-uRPF because it is out of the scope.

Gap Analysis This section elaborates the gap scenarios and key problems of the current operational intra-domain SAV mechanisms.

Intra-domain SAV for Traffic from Sub Networks or Directly-Connected Hosts Towards Goal i described in and shown in , the AS operator can use ACL rules or strict uRPF on appropriate routers to implement intra-domain SAV for traffic originated from a sub network or a directly-connected host. For example, the AS operator can configure an ACL rule on router interfaces facing a sub network or a directly-connected host, containing a set of prefixes which can be used as the source address. By using this ACL rule, the router will block any data packet using a source address out of the set of prefixes. However, the key problem of ACL-based SAV is the high operational overhead. Since the ACL rule is typically maintained manually, the AS operator must update the ACL rule according to prefix changes or topology changes timely. If the AS operator forgets to update the ACL rule when the set of prefixes changes, the outdated ACL rule may improperly block legitimate data packets. Strict uRPF can generate and update SAV rules in an automatic way but it will improperly block legitimate data packets in the scenario of asymmetric routing or hidden prefix. In the following, this section introduces two gap scenarios when using strict uRPF to implement intra-domain SAV.

Asymmetric Routing Asymmetric routing means a packet traverses from a source to a destination in one path and takes a different path when it returns to the source. Asymmetric routing can occur within an AS due to routing policy, traffic engineering, etc. For example, a sub network connected to multiple routers of the AS may need to perform load balancing on incoming traffic, thereby resulting in asymmetric routing. shows an example of asymmetric routing. The sub network owns prefix 2001:db8::/55 and is connected to two routers of the AS, i.e., Router 1 and Router 2. Router 1, Router 2, and Router 3 exchange routing information through the intra-domain routing protocol. For load balancing of traffic flowing to the sub network, the sub network expects the incoming traffic destined for prefix 2001:db8:0::/56 to come from Router 1 and the incoming traffic destined for prefix 2001:db8:0:100::/56 to come from Router 2. To this end, it requires that Router 1 advertises the route information of prefix 2001:db8:0::/56 and Router 2 advertises the routing information of prefix 2001:db8:0:100::/56 through the intra-domain routing protocol. shows the FIB entries of Router 1 and Router 2 associated with the two prefixes.

An example of asymmetric routing While the sub network does not expect traffic destined for prefix 2001:db8:0:100::/56 to come from Router 1, it can send traffic with source addresses of prefix 2001:db8:0:100::/56 to Router 1. As a result, there is asymmetric routing of data packets between the sub network and Router 1. Arrows in the figure indicate the flowing direction of traffic. If Router 1 adopts strict uRPF, by checking the FIB entry that matches prefix 2001:db8:0:100::/56, the SAV rule is that Router 1 only accepts data packets with a source address of 2001:db8:0:100::/56 from Router 3. Therefore, when the sub network sends data packets with a source address of 2001:db8:0:100::/56 to Router 1, strict uRPF on Router 1 will improperly block these legitimate data packets. Similarly, if Router 2 adopts strict uRPF, it will improperly block legitimate data packets from the sub network that use a source address of prefix 2001:db8:0::/56.

Hidden Prefix In the hidden prefix scenario, a host originates data packets using a source address that is hidden or invisible to the intra-domain routing protocol and intra-domain routers. The Content Delivery Networks (CDN) and Direct Server Return (DSR) technology is a representative example of hidden prefix scenario.

Hidden prefix in CDN and DSR scenario For example, in , when the user in AS Z sends a request to the anycast server in AS Y, the anycast server will forwards the request to the edge server in AS X which is close to the user. The edge server in AS X is connected to Router 5 and is reachable only via prefix P2. However, after receiving the request, the edge server will send DSR response packets using the source address of the anycast server (i.e., a P3 address). Since prefix P3 is hidden or invisible to Router 5, DSR response packets originated from the edge server will be improperly blocked if Router 5 uses strict uRPF to implement intra-domain SAV. Specifically, if Router 5 adopts strict uRPF, the SAV rule is that Router 5 only accepts packets with a source address of prefix P2 from the edge server. As a result, when the edge server returns DSR response packets using the source address of prefix P3, DSR response packets will be improperly blocked. In addition, even if Router 5 adopts loose uRPF, it will also improperly block these DSR response packets because prefix P3 does not exist in the FIB of Router 5.

Intra-domain SAV for Traffic from the Rest of the Internet Towards Goal ii described in and shown in , intra-domain SAV is typically deployed on router interfaces facing the rest of the Internet to block data packets using an internal source address (see ). ACL-based SAV is often used for this purpose. The AS operator can configure an ACL rule which contains a set of unacceptable prefixes (e.g., internal prefixes) to block any data packet using a source address of these prefixes. However, the operational overhead of maintaining ACL rules will be extremely high, especially when there are multiple router interfaces that need to configure the ACL rule as shown in .

Intra-domain SAV for Traffic from the Rest of the Internet In addition, loose uRPF can be used in this case to block data packets from the rest of the Internet using a non-global or non-routed source address. But it will improperly permit spoofed data packets using an internal source address because internal prefixes/addresses exist in the router's local FIB.

Problem Statement As analyzed above, the current operational intra-domain SAV mechanisms have significant limitations in terms of automatic update or accurate validation. ACL-based SAV entirely relies on manual maintenance and thus requires high operational overhead in dynamic networks. To guarantee accuracy of ACL-based SAV, AS operators have to manually update the ACL rule in time when the prefix or topology changes. Otherwise, improper block or improper permit problems may appear. Strict uRPF can automatically update SAV rules, but may improperly block legitimate traffic under asymmetric routing scenario or hidden prefix scenario. As analyzed in , it may mistakenly consider a valid incoming interface as invalid, resulting in legitimate data packets being blocked (i.e., improper block problem). Loose uRPF is also an automated SAV mechanism but its SAV rules are overly loose. As analyzed in , any spoofed data packet using a source address covered by the FIB will be accepted by loose uRPF (i.e., improper permit problem). In summary, the AS operator has a comprehensive perspective, so it can configure the correct ACL rules. However, manual maintenance can lead to high operational overhead and improper blocks may occur due to the negligence of the AS operator. uRPF cannot guarantee the accuracy of SAV because it solely uses the router’s local FIB information to determine SAV rules, which may not match the incoming interfaces of legitimate data packets from the source. As a result, strict uRPF will improperly block legitimate data packets in the case of asymmetric routing and hidden prefix, while loose uRPF has limited effect on preventing source address spoofing because it only blocks non-global or non-routed addresses.

Requirements for New SAV Mechanisms To address the problems of current intra-domain SAV mechanisms, this section lists five basic requirements for technical improvements and related discussions that should be considered when designing the new intra-domain SAV mechanism.

Accurate Validation The new intra-domain SAV mechanism MUST improve the accuracy upon existing intra-domain SAV mechanisms. It MUST achieve the two goals described in to block those spoofing traffic from sub networks, hosts, and the rest of the Internet. Meanwhile, it MUST avoid blocking legitimate data packets, especially when there are prefix changes, asymmetric routes, or hidden prefixes. To overcome the improper block problems, routers may need to use more information (e.g., SAV-specific information) besides the local FIB information to determine SAV decisions. By integrating SAV-specific information, routers may learn asymmetric routes or hidden prefixes, resulting in more accurate SAV rules.

Automatic Update The new intra-domain SAV mechanism MUST be able to automatically generate and update SAV rules on routers, rather than relying entirely on manual updates like ACL-based SAV. Even if some necessary initial configurations may be needed to improve the accuracy of SAV, automation helps reduce subsequent maintenance overhead of the AS operator.

Working in Incremental Deployment The new intra-domain SAV mechanism MUST specify the deployment scope (i.e., which routers the mechanism is used on) and MUST provide incremental benefits when incrementally deployed within the specified deployment scope. That is, it MUST NOT be effective only when fully deployed within the deployment scope. In the incremental deployment scenario, it MUST be able to fulfill or partially fulfill the goals described in and MUST avoid improper blocks.

Fast Convergence The new intra-domain SAV mechanism MUST be able to update SAV rules in time when prefix changes, route changes, or topology changes occur in an AS. Two considerations must be taken into account if SAV-specific information is communicated through a protocol. First, the mechanism MUST allow routers to learn the updated SAV-specific information in a timely manner. Second, the mechanism MUST NOT communicate too much SAV-specific information for the SAV function, because this may greatly increase the burden on the control plane of routers and even compromise the performance of the current protocols.

Security The new intra-domain SAV mechanisms MUST NOT introduce additional security vulnerabilities or confusion to the existing intra-domain architectures or protocols. details the security scope and security considerations for the new intra-domain SAV mechanism.

Security Considerations Similar to the security scope of intra-domain routing protocols, intra-domain SAV mechanisms can ensure integrity and authentication of protocol messages that deliver the required SAV-specific information, and consider avoiding unintentional misconfiguration. It is not necessary to provide protection against compromised or malicious intra-domain routers which poison existing control or management plane protocols. Compromised or malicious intra-domain routers may not only affect SAV, but also disrupt the whole intra-domain routing domain. Security mechanisms to prevent these attacks are beyond the capability of intra-domain SAV.

IANA Considerations This document does not request any IANA allocations.

Acknowledgements Many thanks to the valuable comments from: Jared Mauch, Barry Greene, Fang Gao, Kotikalapudi Sriram, Anthony Somerset, Yuanyuan Zhang, Igor Lubashev, Alvaro Retana, Joel Halpern, Aijun Wang, Michael Richardson, Li Chen, Gert Doering, Mingxing Liu, Libin Liu, John O'Brien, Roland Dobbins, Xiangqing Chang, Tony Przygienda, Yingzhen Qu, Changwang Lin, James Guichard, Linda Dunbar, Robert Sparks, Yu Fu, Stephen Farrel etc.