Avoiding Failures The network architect should try to deliver a network that is unaffected by failures and can recover fast in the event of a catastrophe. Core Cisco routers and switches can have dual power supplies and fans, modular chassis-based design as well as additional management components. Redundant modules increase the cost of ownership, but they are typically well worth the investment. It is a good practice for the Core layer switches to have hot-swappable units whenever possible. Hot-swappable units can be installed or detached without first having to turn off the power to the switch. Using these devices decreases repair time and problems to network services. Larger corporations frequently install generators and big UPS devices. These UPS devices stop small power outages from causing big network problems. Decreasing Human Error Human mistakes contribute to network disasters. Unfortunately, adding redundant links and network appliances cannot eradicate these factors. Many network disasters are the outcome of poorly designed networks or inappropriate new equipment. You should never make any change on a production environment before first testing it in a lab environment! Disasters at the Core Layer can cause extensive outages. It is very important to have policies and procedures that describe how modifications are approved, tested and installed. Create a back-out strategy to return the systems to the previous state if changes are not as expected. The appropriate routing protocol for the Core Layer is determined by the scope of the network and the amount of redundant links that are available. A key factor in choosing a routing protocol is how fast it adjusts from a device failure. Convergence Network convergence happens when all routers have comprehensive and precise information about the network. When the convergence time is fast, a network can react very quickly to a change in topology. Elements that influence convergence time include: 1. the speed at which the routing updates replicate in the network, 2. the time that it takes each router to do the calculation to determine the best routes. Picking a Routing Protocol Many dynamic routing protocols provide tolerable convergence times in small networks. In big networks, protocols like RIPv2 converge very slowly and cannot prevent disruption of network services when a link breaks. Usually, in a big enterprise network, EIGRP or OSPF offer the most reliable routing solution. Design Thoughts Many networks use a mixture of dynamic and static routes. Network architects should always consider the amount of routes required to guarantee that all destinations in the network topology are accessible. Long routing tables need substantial time to converge. Finally, the network addressing scheme can seriously affect how well the routing protocol can react to a disaster. Manolis Skoras is a Cisco, Microsoft and HP Certified Trainer and systems-network engineer. Recently he created a How To Pass CCNA exam website to help his students and people around the world to better understand the material they will be tested on, thus having greater success rates. Check Certify4Sure today!