When you launch a high-profile site, Job One for the site’s architects should be making sure it can handle whatever traffic volume comes its way, and that’s especially true for a government jobs web site.
Yet The Wall Street Journal reportsthat in the two weeks since a federal jobs web site, USAJobs.gov, relaunched it’s been hampered by performance issues. The embarrassment for the government is compounded by the fact it let Monster.com run it for a number of years without a problem.
When the Feds took over, and undertook a site redesign, they obviously didn’t do enough pre-launch performance testing and the site has been having issues since.
How bad is it? Well, according to The Washington Post, the Office of Personnel Management (OPM) which oversees the site has been sending out daily updates trying to convince everyone the problems are getting better. That may be so, but initially the Wall Street Journal article reported, “The volume of requests instantly crushed government servers, slowed the system and locked out thousands of applicants.”
When a high profile site crashes these days, it instantly becomes fodder for the social networks, which just makes the government site architects look all that much more unprepared. In this case, the site had its own Facebook page and people were, shall we say, less than enthused about the crash.
As we reported, a similar situation happened in September when Target took over its own web site management from Amazon Web Services. When the web site launched a new low-cost designer clothing line from Italian designer Missoni, the site crashed and as with the USAJbos site, people took quickly to Facebook and Twitter to complain.
Federal Computer Week reports that OPM did a good job initially communicating with users on Facebook and Twitter, but then began to go silent, leaving users to stew even more. FCW wrote that Federal officials eventually blamed the performance issues on volumes that were 5 times higher than previous peak times.
In addition to performance issues, users also reported problems with the site’s search tool, meaning when people finally got into the site, they couldn’t access the jobs they wanted through the search engine.
While you could probably excuse the performance issue as traffic spike that went far beyond any previous traffic, not testing and tuning the search engine is a bigger problem.
And all of these problems can be traced to conducting adequate testing prior to a site launch. If you do load testing to see if the site can handle unusually high traffic, you should be OK. If you do some random, common searches to make sure your site search engine is returning reasonable results, you should be fine.
Without inside knowledge of the development process, it’s hard to know exactly what happened here, but it sounds like in this case, the site architect’s underestimated the site’s popularity and didn’t do adequate testing up front to make sure there would be a minimum of glitches on launch day and beyond.
And every site architect, regardless of where they work, should be confident the site will run well on Day One or face the kind of scrutiny this one has when things go awry.