Saturday, October 25, 2014

Test, Rinse, Repeat

With the tens of thousands of lines of code that go into modern software and the inconsistencies that can sometimes occur when programs are compiled to machine language, it can be very hard to guarantee the integrity of almost any software. For organizations that rely on the quality, reliability, and security of their software, no hole can go unpatched. To achieve this, rigorous software testing is an absolute necessity.
One of the most difficult situations any piece of software can undergo is extreme load. With the ubiquity of the Internet and sheer computing capacity computers have today, situations of extreme load are no longer very uncommon. Denial-of-service attacks happen almost on a day-to-day basis, and software bottlenecks can occur when poorly optimized code is fed so much data too quickly.

The problem with software operating at extreme load is that it often doesn’t reveal its most serious problems until it is subjected to such high loads. Instead, it will function normally under normal conditions and then usually fail dramatically when under stress—sometimes with serious security consequences.

Software stress testing is the only way to guarantee that a piece of software will not fail under extreme load, or that it will fail gracefully and not reveal its secrets when it does so. Given the plethora of possible scenarios, the only way to stress test software reliability is to keep going at it from every angle, analyzing each result along the way: test, rinse, repeat.

No comments:

Post a Comment