]> the extensions of BGP-LS to carry security capabilities China Mobile
BeiJing China chenmeiling@chinamobile.com
China Mobile
BeiJing China suli@chinamobile.com
Security Internet Engineering Task Force Internet-Draft keyword2 As users' traffic faces more unpredictable attacks during transmission, there are more and more end-users now need high security data transmission guarantee, they need ISPs to provide security protection capabilities on the data forwarding path, but it is very difficult for operators to manage the security attributes of nodes through control surfaces. ISPs need to have real-time awareness of the security capabilities available in the network, then form a security capability map, finally provide security protection for users at the routing level. The goal of this draft is to collect the security capabilities of nodes, which will be one of the factors to form the routing topology, and use the routing programming capabilities to form a secure routing path. The security capability includes healthy information(such as the device software is up-to-date), security service information, device information(such as the manufacturer information of the equipment). The BGP-LS protocol is extended to carry the security capabilities of the node. The controller collects topology information, forms a topology path with security capabilities according to security requirements, and supports SRv6 path sending to execute node forwarding through programming.
Introduction SRv6 (Segment Routing IPv6, IPv6 segment routing) is based on source routing and centralized routing. It can realize network intelligent programming and select forwarding paths according to customer needs. At present, there is a lack of effective technical means to inject security factors into the process of collecting network topology and centralized routing to achieve safe routing path forwarding. The most important reason for using BGP-LS as the extended basic protocol is that BGP-LS shields the differences of other routing protocols, and the underlying routing protocol types do not need to be considered when transmitting security capabilities. RFC7752 standardized North-Bound Distribution of Link-State and Traffic Engineering (TE) Information Using BGP, describes a mechanism by which link-state and TE information can be collected from networks and shared with external components using the BGP routing protocol, using a new BGP Network Layer Reachability Information (NLRI) encoding format. BGP-LS is a new way to collect network topology. The topology information discovered by the IGP protocol is summarized by the BGP protocol and sent to the upper controller. With the powerful routing and routing capabilities of the BGP protocol, there are three types of BGP-LS routes, which are used to carry node, link and route prefix information respectively. The three routes cooperate with each other to complete the transmission of topology information. The node routing function is to record the node information of the topology, the link routing function is to record the link information between two devices, and the address prefix routing function is to record the network segment information that the node can reach. The state information NLRI collected by BGP-LS is described in TLV (type/length/value triplet) format. Each link state described by NLRI can identify a node, link or prefix. Therefore, three types of NLRI are newly set in the standard, of which type 3 and 4 are used to distinguish the prefix of IPv4 and IPv6. There are only two types of NLRI attributes in the original BGP protocol: MP_ REACH_ NLRI, attribute type 14; MP_ UNREACH_ NLRI, attribute type 15.
BGP-LS node type carries security capability
Collection model of security capabilities
New Node Attribute TLVs The Local Node Descriptors TLV contains Node Descriptors for the node anchoring the local end of the link. This is a mandatory TLV in all three types of NLRIs (node, link, and prefix). Node attribute TLVs are the TLVs that may be encoded in the BGP-LS attribute with a Node NLRI. The following Node Attribute TLVs are defined: The security capability is transferred by adding the security capability attribute to the attributes of the local node.
Usage of new attribute When programming the routing path, take the security capability requirement as one of the inputs. The description of the security capability requirement can be structured or one-dimensional matrix, which only needs to be consistent with the router's security capability description; There are many routing rules. After introducing security capability requirements, it is necessary to dynamically adjust the security capability as the position of routing rules according to the requirements. The main rule strategies are: ① Select the routing node that meets the security requirements as the forwarding node when the path is reachable; ② Select the shortest path when all the safety requirements are met; ③ When the same path length and security requirements are met, select the path with small load for forwarding.
BGP-LS Link type carries security capability
Collection model of security capabilities The router and its attached security products are the basic units. When collecting status information, only some nodes can directly transmit the node status information to the controller through the BGP-LS protocol. Other nodes that do not directly transmit the node information need to transmit the node information to the direct node to achieve the transmission of security capability information. Therefore, for non direct nodes, It is required to report its own security capability information through the BGP-LS link state data packet.
New Link Attribute TLVs The new attribute describes the link security capability and transmits the link security capability information through this attribute.
Useage of new attribute | | SCing B | <----------------+ |SCing | +----------------> | | Figure 9: Assosiation security capability interaction ]]> The Assosiation security capability depends on the security capability enabled by the node. As a node directly connected to the controller, node B first interacts with the enabled security capability information of the opposite end in a two-way manner, and then the opposite end initiates the transmission of the assosiation security capability information. The decision of assosiation security capability can be divided into two situations: one is under the same security domain, and the other is under different security domains. 1. The decision rules for link security capabilities under different security domains are as follows: SCing represents the enabled security capabilities of a node. Example: SCing A=[1,0,0,1,0,……]，SCing B=[1,1,0,1,0,……]，SCing Assosiation<A,B> = SCing A && SCing B When the link passes through more than two nodes, it is necessary to logically and operate the security capabilities of all nodes in the path to obtain the link security capabilities.
  1. The decision rules of assosiation security capability in the same security domain are as follows: SCing indicates the security capability of a node that has been enabled. Example: SCing A=[1,0,0,1,0,……]，SCing B=[1,1,0,1,0,……]，SCing Assosiation<A,B> = SCing A || SCing B
When the link passes through more than two nodes, it is necessary to logically or operate the security capabilities of all nodes in the path to obtain the link security capabilities.
BGP-LS Prefix type carries security capability
Collection model of security capabilities The router and its attached security products are the basic units. When collecting the status information, only some nodes can directly transmit the node status information to the controller through the BGP-LS protocol. Other nodes that do not directly transmit the node information need to transmit the node information to the directly connected node to achieve the transmission of security capability information. In the figure, nodes A and E are direct connected nodes, which are connected to their respective controllers. Nodes A and E are responsible for collecting the security capabilities of other nodes in their respective fields.
New Link Attribute TLVs The IPv4 and IPv6 Prefix NLRIs (NLRI Type = 3 and Type = 4) use the same format, as shown in the following figure. An AS has at least one super direct connection node, which has the security capability information of all nodes under the AS. By adding new attributes to Prefix, the security capabilities of the entire AS can be transferred. AS Security capabilities means the security capability information of all nodes under the AS, that is, the security capability information of all nodes is spliced, such as {[IP address (A)+node security capability], [IP address (B)+node security capability]...}.
Usage of new attribute
IANA Considerations This memo includes no request to IANA.
Security Considerations TBD