SRLG stands for Shared Risk Link Group. It defines a concept that multiple different services may suffer from a common network failure if they are sharing a common failure risk. Such risks can be a fiber link, a man-hole, an operational center, etc. The consequence of the failure of the common risk is that all the services that share the risk would be affected or even totally interrupted.
The examples of SRLGs include
Figures below show the above three SRLG failure scenarios, including (a) common link (or conduit) failure, (b) common node (or operational center) failure, and (c) subnetwork failure.
In addition, the SRLG concept is network layer-related. A link or node that is an SRLG for its upper layer services can also share an SRLG in a lower layer. For example, a wavelength channel is an SRLG for all the STM-n tributaries that it carries. Meanwhile, this wavelength channel shares a fiber with all the other wavelengths contained within the fiber; the fiber is a common SRLG of all these wavelengths.
The layered SRLG relationship is transferred from upper layers to lower layers. Specifically, if B is an SRLG of A, and C is an SRLG of B, then C is an SRLG for A. A fiber is an SRLG of a lightpath channel, which is by default an SRLG of an STM-n tributary that is contained within the lightpath.
Control plane SRLG
In addition, a transport network is often divided into a data plane and a control plane. The control plane provides control functionality to administrate the data plane. The failure of the control plane would lose the control of the whole data plane; thus, control plane is also considered as a type of SRLG for all the services carried on the data plane.
Protection from SRLG failure
To enable network services to survive from SRLG failures, many survivable network design approaches have been developed. A common key principle for SRLG protection in these approaches is to set up a pair of SRLG-disjoint routes respectively for the working and protection paths. Figure below shows pairs of disjoint routes to protect services from different SRLG failures. For the link SRLG, we can find a pair of working and protection routes that do not share any common links; for the node SRLG, we can find a pair of routes that do not share any common nodes, and for the subnetwork SRLG, we can ensure that a pair of working and protection routes do not traverse any common subnetworks.