Based on Identifier Network, we propose a solution for traffic engineering, which can be divided into two distinct parts: End-to-End traffic engineering and Neighbor-to-Neighbor traffic engineering. For each scenario, we develop a routing decision method for both routers and other network entities, such as IDMS (Identifier Mapping Server in Identifier Network). To analyze the feasibility of the solution, we collect Routeviews data set and the results show that the scheme proposed could reduce the burden of the core routing table.
TopIntroduction
With the rapid development of Internet, today’s Internet has become the center of the global information, resource, application and service. However, the Internet is facing serious scaling problems (Xiaoliang, 2010; Meyer, 2007). In 2007, there were about 200,000 IP address prefixes in the routing tables of the core Internet Routers (Huston, 2012). Until 2012, the routing entries have exceeded 400,000 prefixes for IPv4 and almost 11,000 prefixes for IPv6, and the global routing table size has experienced an increase in a nonlinear rate (Huston, 2012; Elmokashfi, 2010). Regarding the working report of Internet Assigned Numbers Authority (IANA) (Meyer, 2007), the reasons contributing to the fast explosion of the global routing table mainly include the following: multi-homing, inbound Traffic Engineering (TE), non-aggregated address allocations (a big portion of which is inherited from historical allocations), and business events such as mergers and acquisitions.
To address this routing scalability problem, many organizations including IETF (Internet Engineering Task Force) LISP (Locator/ID Separation Protocol) Working Group (LISP, 2012) and IRTF (Internet Research Task Force) RRG (Routing Research Group) (RRG, 2012) have discussed the idea of separating the node’s identity from its topological location. Identifier Network, as a novel ID/Loc (Identifier/Locator) separation architecture, separates the identifier and locator roles of IP addresses into two evolving spaces: Accessing Identifier (AID) and Routing Identifier (RID).
The AID indicates a host’s identity and is applied in the transport and application layers, whereas the RID indicates the current topological location of a host and is used in the network layer. In addition, Access Router (AR) deploys on the edge of the core network and provides the mapping functionalities between the AID and RID.
The ID/Loc separation solution not only can reduce the routing table size in the Default Free Zone (DFZ) (Quoitin, 2007) and natively support mobility (Zhang, 2009; Luo, 2011), but also provides promising opportunities to reconsider routing optimization for inter-domain (inbound) traffic engineering.
In this chapter, we propose a solution for implementing Traffic Engineering in Identifier Network. The method contains two distinct parts: End-to-End traffic engineering and Neighbor-to-Neighbor traffic engineering. For each scenario, we develop a routing decision scheme for both routers and other network entities, such as IDMS (Identifier Mapping Server in Identifier Network) (Luo, 2011). The aim of both the above proposed solutions is to reduce the burden of the core routing tables.