Abstract
Survivability is the ability of the network to withstand faults and attacks including equipment and link failures. The main goal of survivable network is to be able to perform fast recovering at as small cost as possible (i.e., using minimum resources). As wavelength routing network paves the way for network throughputs of possibly hundreds of Tb/s, network survivability assumes critical importance. A short network outage can lead to data losses of the order of several gigabits. Hence, protection, or dedicating spare resources in anticipation of faults, and rapid restoration of traffic upon detection of a fault are becoming increasingly important. In this paper, we will explore the state of the art in network survivability and corresponding network control issues.
Key Terms in this Chapter
Lightpath: Light path is a path from the source to the destination in optical domain, meaning data remains in optical domain without any optical-electronic- optical conversion in between the source and the destination.
Shared Risk Group (SRG): A group of links where all the links in the group has faults at the same time. For example, a group of links connecting different nodes may be deployed such that they are going through the same conduit at some point. Then all links passing the conduit will have faults when the conduit has been cut.
Wavelength Division Multiplexing (WDM): A way to utilize the huge bandwidth by dividing the bandwidth into multiple channels called wavelength and operating channels in parallel. Theoretically, a single fibre can carry up to 30 Tera bits per second, meaning we can deploy 3000 channels where each channel can carry 10 Giga bit per second. Considering the fact that no electronic device matches with the speed of the optical fibre’s data transfer speed, WDM is the only multiplexing method to fully utilize the bandwidth that optical fibre can provide.
Link Protection: Link protection is a link-based protection scheme in optical networks, meaning each link has a protection path it is associated with. Protection is initiated by the end nodes of the fault link, and all the paths over the fault link will be rerouted through the protection path of the fault link. Therefore, link protection takes less protection time than path protection at the cost of fine granularity.
Restoration: Restoration does not reserve any resource in advance, rather than it computes a new path from the source to the destination when the existing path becomes unavailable. Therefore, restoration utilizes resources more efficiently than the protection. However, it has much delay associated to compute a new path on the fly when fault occurs.
Path Protection: Path protection is a path based protection scheme, meaning each path has an edgedisjoint protection path associated with it. Protection is initiated by the source node of each path, and each path over the fault link will be rerouted through its own associated protection path. Therefore, path protection takes more protection time than link protection, and provides protection with fine granularity.
Multi-Protocol Label Switch (MPLS): A combined control plane for networks. It allows using switch level (layer 2) forwarding and IP router level (Layer 3) routing. In follows, it not only improves speeds of networks but also provides scalability and flexibility enough to do traffic engineering.
Routing and Wavelength Assignment (RWA): Means a way to find a route and to select wavelength for the route from the source to the destination. The complexity of the RWA is caused by the wavelength continuity constraint, meaning all the links belong to a route from the source to the destination have to have same wavelength. RWA is most heavily studied in WDM optical network since the network performance in terms of blocking probability, wavelength utilization and etc. is heavily depend on it.