Abstract
Techniques are described herein for a two-fold process where a first Machine Learning (ML) model is engaged in building a Dependency Mapping Table (DMT) based on the data from the network (network observations). An Interior Gateway Protocol (IGP) database, service graph information, and in-band Operations, Administration and Management (iOAM) data is leveraged to feed into the ML model to build the DMT. The DMT is built with a per physical element precision (fiber optical cable in each direction). In other words, the database is built in a way to enable identification of the list of (connected/non-connected) services and protocols that potentially relies on the fiber cable. In the second fold, the ML model uses the dependency mapping built in the previous phase to identify the potential cause for an issue and prioritize the relevant alarms. In addition, it may be tied together with existing failure prediction mechanisms for any layer that in turn will be used with the above database to prioritize the alarm/notification to an operator Operations Support System (OSS) and take any necessary pre-emptive action on the above layers.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Pignataro, Carlos M.; Nainar, Nagendra Kumar; and Eguiarte, Alejandro, "SERVICE LAYER DEPENDENCY MAPPING BASED HEATMAP FOR FAULT ISOLATION", Technical Disclosure Commons, (January 16, 2019)
https://www.tdcommons.org/dpubs_series/1888