19-09-2017, 03:27 PM
As the Internet occupies an increasingly central place in our communications infrastructure, the slow convergence of routing protocols after a network failure becomes a growing problem. MRC is strictly offline, and supposes only destination-based hop-by-skip forwarding. MRC relies on maintaining additional routing information on the routers and allows packet forwarding to continue on an alternate output link immediately after a fault is detected. It can be implemented with only minor changes to existing solutions. In this paper we present MRC, and analyze its performance with respect to scalability, backup path lengths and load distribution after a failure. It also shows how an estimate of traffic demands on the network can be used to improve the distribution of recovered traffic, and thus reduce the chances of congestion when MRC is used.
In recent years the Internet has been transformed from a special purpose network into a ubiquitous platform for a wide range of daily communication services. The demands for reliability and availability of the Internet have increased accordingly. An interruption of a link in central parts of a network has the potential to affect hundreds of thousands of telephone conversations or TCP connections, with obvious adverse effects. The ability to recover from failures has always been a central goal of Internet design. IP networks are intrinsically robust, since IGP routing protocols such as OSPF are designed to update forwarding information based on the modified topology after a failure. This re-convergence assumes the complete distribution of the new link state to all the routers in the network domain. When the new status information is distributed, each router individually calculates new valid routing tables. This network IP network convergence is a time consuming process and a link failure or node is usually followed by a period of routing instability. During this period, packets may fall due to invalid routes. This phenomenon has been studied both in the IGP and BGP context, and has an adverse effect on real-time applications. Events that lead to re-convergence have been shown to occur frequently. The PGI convergence process is slow because it is reactive and global. Reacts to a bug after it has occurred, and involves all routers in the domain. In this article we present a new scheme for handling link and node errors in IP networks. Multiple Output Configurations (MRC) is a proactive, local protection mechanism that allows recovery in the millisecond range. The MRC allows packet forwarding to continue over the next preconfigured jumps immediately after fault detection. Using MRC as a first line of defense against network failures, the normal IP convergence process can be put on hold. This process then starts only as a consequence of non-transient faults. Since no global rerouting is performed, fast fault detection mechanisms such as hardware alerts or quick alerts can be used to activate MRC without compromising network stability. MRC guarantees the recovery of any single link or node failure, which constitutes a great majority of the faults experienced in a network. MRC makes no assumptions regarding the root cause of the failure, for example, if packet forwarding is interrupted because of a failed link or a failed router.
In recent years the Internet has been transformed from a special purpose network into a ubiquitous platform for a wide range of daily communication services. The demands for reliability and availability of the Internet have increased accordingly. An interruption of a link in central parts of a network has the potential to affect hundreds of thousands of telephone conversations or TCP connections, with obvious adverse effects. The ability to recover from failures has always been a central goal of Internet design. IP networks are intrinsically robust, since IGP routing protocols such as OSPF are designed to update forwarding information based on the modified topology after a failure. This re-convergence assumes the complete distribution of the new link state to all the routers in the network domain. When the new status information is distributed, each router individually calculates new valid routing tables. This network IP network convergence is a time consuming process and a link failure or node is usually followed by a period of routing instability. During this period, packets may fall due to invalid routes. This phenomenon has been studied both in the IGP and BGP context, and has an adverse effect on real-time applications. Events that lead to re-convergence have been shown to occur frequently. The PGI convergence process is slow because it is reactive and global. Reacts to a bug after it has occurred, and involves all routers in the domain. In this article we present a new scheme for handling link and node errors in IP networks. Multiple Output Configurations (MRC) is a proactive, local protection mechanism that allows recovery in the millisecond range. The MRC allows packet forwarding to continue over the next preconfigured jumps immediately after fault detection. Using MRC as a first line of defense against network failures, the normal IP convergence process can be put on hold. This process then starts only as a consequence of non-transient faults. Since no global rerouting is performed, fast fault detection mechanisms such as hardware alerts or quick alerts can be used to activate MRC without compromising network stability. MRC guarantees the recovery of any single link or node failure, which constitutes a great majority of the faults experienced in a network. MRC makes no assumptions regarding the root cause of the failure, for example, if packet forwarding is interrupted because of a failed link or a failed router.