Fault Tolerant Management

Last October I was involved in co-ordinating a FP7 Call 5 research project proposal. The primary objective of the project was to research mechanisms for engineering stable and reliable self-* management systems - an essential requirement for the adoption and widescale deployment of automated network management systems. Typically, carrier-grade networking equipment is designed to be fault-tolerant meeting high availability requirements, for example 99.999% available or just over 5 minutes of downtime per year. The management systems, on the otherhand are not - the theory being, that an outage on the management system should not prevent the network from continuing processing calls or traffic. Even if the management systems are highly-available themselves, the act of making network configuration changes can get quite complex when taking into account everything that could possibly go wrong. Network configuration changes, consequently, tend to be meticulously planned and then pushed into the network in a controlled manner to minimise risk of performance degradation or an outage.

The business driver for self-* management is to reduce the cost of running the network. However, if the management system is to be capable of self-* management then it must also be capable of functioning in a reliable and stable manner, or in otherwords, it must be Fault Tolerant. The management system is required to automatically detect events in the network, identify what counter-measures must be taken to react by pushing out configuration adjustments to ensure optimal network performance. It needs to do so reliably in order to ensure high-availability. Without achieving this, the step towards widescale adoption of self-* management might prove to be a bridge too far.

Webinar
Attended the following webinar this week, End-to-End Monitoring Cuts Churn, Lowers Opex from Harmonic and Sencore. For IPTV services provided by MSOs need multipoint monitoring for video distributed for verifying the quality of the signal through the network and gather realtime performance data. Monitoring should not limited to IP layer only. The main objective is to identify problems and resolve them before the custormer notices!!!