Answers to your Questions

What is the cause of most network outages?

Most problems occur during network upgrades.

Many examples of outages being caused by network software upgrades can be found on the web. In one such article, Jeff Moore, an analyst at Current Analysis Inc. in Sterling, Va., claims, "Most network switch problems occur during software changes." One of AT&T's worst outages involved its frame-relay network, which went down for 26 hours in April, 1998 during a software upgrade.

Today, most carriers and Internet Service Providers run a suite of regression tests within their test labs before new network software is released into the live network. Although it's true that equipment vendors run their own system and Quality Assurance tests, Service Providers run their own regression tests to ensure that the new software will interoperate with equipment from other vendors, will work in the Service Provider's unique network configuration, and will not significantly reduce baseline network performance.

A simple approach to regression testing is to run off-the-shelf scripts, such as conformance test suites and automated test applications, augmented with a barrage of manually executed test cases. A detailed Test Plan is developed to enable test engineers to configure and run tests quickly.

For more rigorous and repeatable testing, a Service Provider would develop its own test scripts or customize off-the-shelf applications. Automated regression testing is a large investment that ultimately pays off through reduced testing cycles, faster deployment of new service capabilities, and reduced risk of network outages and service downtime.

For more detailed information on network outages, see Preventing Network Service Outages.