IT has been in the news lately with the less-than-successful launch of the government healthcare exchange website, healthcare.gov. For some, it is an indictment of technology gone wild and for others it is clear evidence of political ineptitude. I am not used to seeing political cartoons about information professionals, but it is becoming the new norm. The reasons behind the failure are complex, but I would like to focus on one this week that I think will help all of us to avoid missteps such as this in the future. One thing that is imperative at a new launch is load testing.
Load testing is defined as executing the largest number of tasks, under test, that the system can handle. It also means understanding the behavior of the system under that maximum load. Is it just slow or is it completely unavailable? One may be acceptable and the other unacceptable. Load testing is most successful when the maximum number of users or concurrent processes is known in advance. In the case of the healthcare exchange, I believe that there is enough data to predict how many people would try to access the site in any given period.
All the Way Down the Line
It all sounds so simple but it can get very complicated. Not only do you need to test the potential load on the website and the web application, but you also need to test the potential load on the web server, the database serving the information, the database server, and the network tying it all together. Weak performance in any of these can cause the kind of problems seen with healthcare.gov. This is where an IT troubleshooter is worth their weight in gold. Someone who understands the interoperability between all of these systems and processes can root out potential problems before the application goes live.
When to launch a new application or website is also partly common sense. Computer testing can only go so far. If, what is reported in the Washington Post is correct, not only did the load test fail, the common sense test failed as well:
“Days before the launch of President Obama’s online health insurance marketplace, government officials and contractors tested a key part of the Web site to see whether it could handle tens of thousands of consumers at the same time. It crashed after a simulation in which just a few hundred people tried to log on simultaneously. Despite the failed test, federal health officials plowed ahead.”
In cases such as this I think of the immortal words of Walt Kelly: “We have met the enemy, and he is us.”
It is possible to correctly predict the performance of an application and have a wildly successful launch. Do you have stories of successes, large or small? Do you have stories of failures that you would just as soon forget but provided great lessons to you and others? I encourage you to share your story so that we can all learn. Let me know your thoughts.
Kelly Brown is an IT professional, adjunct faculty for the University of Oregon, and academic director of the UO Applied Information Management Master’s Degree Program. He writes about IT and business topics that keep him up at night.