It’s this quote that makes me question whether you really understand launch day for a highly anticipated web 2.0 website. The precise volume really does matter and what matters most in this situation is that the metric you are talking about: number of users who will qualify. In the case of Obamacare, the people visiting the site were from across the spectrum: People who knew they’d qualify, people who knew they wouldn’t qualify, people who were unsure if they would qualify. All of these people descended on one website at around the same time on the same day. This unnatural spike in visits is the reason servers chuck a skitz and stop working.
I also imagine that if you were working as a developer then you probably didn’t have much to do with load testing what you developed. As I said above: even robust load testing prior to launch is only an indication of expected performance. Real-world users don’t follow rules like simulated ones. Some food for thought: Twitter goes Fail Whale once in a while (less and less lately) and all they’re ever dealing with is adding 140 characters at a time to a database.
