Design and analysis of high availability networks

This study provides an analysis of representative critical service networks based upon research of component failure data. We used IP-SURVIV, a Noblis survivability analysis tool, to assess the availability of three representative critical service networks as components were disabled. Our findings indicate that the generally accepted high availability network topology and routing strategies - dual-homing access nodes and providing a primary and a single alternate route between all backbone node pairs - may result in surprisingly low network availability. We present architecture guidelines which significantly improve network availability. In light of our findings, entities requiring robust, high availability network service should consider reassessing the availability of their networks and also consider implementing both a multi-carrier, 3- connected backbone and a second MPLS backup route to achieve high availability (i.e., ges 0.9999).