One truism about today’s IP networks is that complexity is growing. Given this, a key question that should be asked by managers responsible for large IP networks is whether their network management is keeping pace with network complexity. In too many cases, the answer is no. Most network management tools only understand the network in discrete, device-level pieces, and “end-to-end” performance management tools don’t really look at network behavior, rather they monitor the symptomatic effects of network problems on end-user and service-level performance metrics.
What does network complexity mean? One simple way to look at it is that when there is a high degree of redundant links between various devices in the network, there are a correspondingly high number of alternate paths that traffic can take between any two points in the network. This corresponds to what can be called Eslambolchi’s Law of Telecom Complexity:
Telecom complexity, C, is calculated roughly as follows:
C ∝ (Number of protocol domains) X
(Number of networks per domain) X
(Avg. number of inter-network interfaces per network)
• Over time, T, the number of protocol domains increases proportional to T.
• The number of networks per domain increases proportional to T.
• The number of inter-network interfaces per network is proportional to the number of networks per domain.
Hence, C ∝ T^3
The problem is that these variable paths in increasingly complex IP networks can change at any given moment, based on dynamic recalculations performed by routers. This sort of variability and unpredictability is the enemy of cost-effective network operations. For example, troubleshooting takes much longer because it is very difficult and many times impossible to localize which part of the network to look at. Especially when problem analysis is done after the symptoms stop occurring, hours can be wasted analyzing the wrong part of the network. Not only is there a direct operations cost to lengthened troubleshooting processes, but when root causes can’t be identified, the underlying issues accumulate over time and impact network and service quality. Given that perceptions of service quality are one of the most important differentiators that customers apply to their service provider this is clearly an undesirable cycle to get stuck in.
For a long time, the complex routing dynamics in IP networks were considered too difficult to understand, but that has changed with the advent and now fairly wide-spread adoption of IP route analytics technology. The key innovation in route analytics is using the network’s live routing protocols as a source of real-time network management information. By recording and processing the network’s continuously updated routing information, route analytics tools create and maintain an extremely detailed, always accurate network topology which can be used interactively to analyze, troubleshoot and even model network dynamics.
Hundreds of service providers, mobile operators, cable MSOs, as well as large enterprises and government agencies are now doing just that by making their network management more intelligent with route analytics. Route analytics recorded history of routing and traffic dynamics can be used to analyze the actual paths of troubled service traffic for faster troubleshooting, rather than hunting and guessing, saving significantly on operations costs. Modeling and simulation capabilities of route analytics can be used to accurately understand how proposed routing and traffic changes would impact existing service traffic, and prevent costly service disruptions, before making changes. Route analytics can also be used to systematically audit carrier-sized networks to inventory out-of-policy routing conditions, identify vulnerable points in network architecture where a failure would cause a significant, negative impact on traffic, and find unused links and routers—expensive assets that can be profitably redeployed. The operational and capital savings from such intelligence can be truly significant.
In these days of economic constraints, it’s imperative to work smarter. Adding routing intelligence is a smart way to increase the leverage that network management tools and processes provide to operate networks and deliver services profitably.
