* Router misconfigs are often downtime culprit
Performance management products abound and we regularly discuss them here. In general, these products look at network performance in a “steady state.” That is, they are designed to measure performance for networks that are configured correctly. At that point, the performance-related issues come down to variation in traffic load, traffic types and application performance.
But what happens when networks are not properly configured? We are accustomed to hearing that the primary cause of failures is a physical circuit break. We don’t dispute that, and it is indeed the primary cause at the transport layers.
However, a recent study by the University of Michigan and Sprint, as we learned in a recent briefing from route-analytics company Packet Design, showed that more than half of the problems causing downtime in IP networks – 59% – pertain to routing management issues.
Digging a bit more deeply, 36% of these problems are attributable to router misconfigurations, and 23% come from a category broadly described as “IP routing failures.” By contrast, of the remaining 41% of problems, link failures of some form account for 32%, and “other causes” comprise the remaining 9%, according to the study.
Those familiar with the relatively nailed-up world of frame relay and ATM will find this to be a departure. In a traditional frame/ATM PVC environment, the forwarding tables that tell various network switches where to send traffic and when are relatively static. By contrast, in an IP-centric network, the reachability information passed from control-plane protocols is much more dynamic, so there is much more opportunity for configuration mistakes to be made.
In a world of “private IP” where the enterprise uses more traditional services like ATM, frame relay, and even private lines between routers, the impact of misconfiguration is annoying but seldom catastrophic. The data that is intended to stay company-confidential remains within the organization.
However, in an IP-based MPLS VPN, a misconfiguration can cause a breach of security. Next time we’ll explain why.
