Xbox Live officially launched November 14, 2002. The private launch party in Hollywood featured celebrities like Samuel L. Jackson and Freddy Prinze Jr. But the real party was online, where players all over the world were discovering the service for the first time. They sold 150,000 Xbox Live Starter Kits in the first week alone.
Hours into the launch, Multerer was told there was a problem. The numbers didn’t look right.
“We turned the service on and then went into the cafeteria to watch the numbers. It had a party atmosphere, but only lasted a few hours as the numbers didn’t make sense and there was clearly something wrong. It turned out to be a bug in the software that reported the numbers, not the service itself. So no big deal, but it did interrupt the party.” No big deal, except that once they fixed the bug it became clear that they had shattered their projections.
In many industry launch stories, the servers are overwhelmed by the initial, and unexpectedly massive, success. Breakdowns on launch were not uncommon. The original Pong coin-op test at Andy Capps famously stopped working, causing all kinds of consternation at Atari, but it turned out that the coin box was so stuffed that it couldn’t take any more quarters. Launches of games like Ultima Online and EverQuest featured overloaded servers and last-minute emergency measures. It was a common story, but not with Xbox Live. Even though the launch was far more successful than they had anticipated, they had come prepared. In fact, considerably over prepared.
Thomason reflects on what happened. “We overbuilt the data centers—maybe for the first time in history, and maybe since. We actually ended up shutting down some of the data centers later because we had more capacity than we needed. It ended up ramping up really well, but we were just super paranoid. We didn’t think we could live through an experience where we didn’t have enough capacity for people playing. We wanted a console experience for Live, meaning it had to always work, it had to be fast, and all that stuff. The Xbox Live launch went extraordinarily smoothly, and people just came up and played, and just played… and it worked.”
Rolling the Dice
As impressive as the stability of the initial launch was, perhaps even more impressive was how they planned it. Multerer revealed to me, he says for the first time ever, the very technical process they implemented to determine how many machines they would need to handle the anticipated load. “As we were rolling through the summer before launch, MarkV (VanAntwerp) and I had to provide capacity planning numbers for the services. Very important as if you buy too few machines/services then you can’t handle the load when the service turns on, but if you buy too much then you blow the budget.
“The problem was that there was nothing similar to model it against. So we built a big spreadsheet, came up with the best and worst case numbers. Then I got out my old Dragon Dice from Dungeons and Dragons and we literally chose what felt like the right curves and rolled the dice to enter the numbers.
“‘This one feels like 20tps + 3d12’ (tps = Transactions per Second) (3d12 = A twelve sided die, rolled 3 times and added together) We put that nomenclature in the spreadsheet so we could remember what we did, but never told anyone about it (until now) because the ops and test teams wouldn’t have liked the randomness of it. In reality it wasn’t very random and was actually a good way to get probability curves for the planning. Worked out just fine.”
January 01, 2002:
Microsoft introduced Xbox LIVE in Nov. 2002, describing it as the first comprehensive, online game arena fully dedicated to fast-action broadband gaming experiences. The company sold 150,000 Xbox Live Starter Kits in the service’s first week.
Better Strategy
It was huge. They came up with a better strategy than Sony did, and Sony Corporation was going through a lot of turmoil at the time, and they weren’t well led, and they let Microsoft win.
-Brad Silverberg
More Comments about Xbox Live
Xbox Live brought the Internet to consoles in a real and substantial way. Of course it offered game play connectivity, but it also created a community, a medium for communication and a completely different way to distribute games.
I would say that the success of Xbox Live forced everyone else to respond with their own services. Which were successful to varying degrees. Overall, I think Xbox Live was viewed as the best of these offerings and was probably the key differentiating feature for both the original Xbox and the Xbox 360.
-Stuart Moulder
We launched it on the day we planned. Our security stuff has always worked very well. It was all designed very early on, and they’re still using a similar scheme today, all these years later. As I understand it now from talking to Marc, Xbox Live is the most profitable part of Xbox.
Xbox Live never paid back much of the loss from Xbox. It was profitable to the tune of maybe $200 million, but we’re talking about losses of billions for Xbox. The general number that’s thrown around is about $4.5 billion for the original Xbox. It’s a staggering amount of money. But to put it in perspective, I think Microsoft lost very close to that same amount of money with the debacle of Windows Longhorn, how it evolved into Windows Vista. I think we flushed at least $3 billion doing that. That’s a different story, but Microsoft has definitely lost lots of money over the years.
-Jon Thomason
We built a successful, entertainment-based social service before social services were cool. A lot of people they go for a date, they go for reduce your risk, they go for shipping, but it’s not interesting to reduce your risk so much that the product you produce doesn’t have enough value. You’ve got to take enough risk, and you’ve got to do something big enough that success is worth it. Like I said in the beginning—and this is super important to me—I do not want to build any projects that would be successful on the day we start. I want to build products that will be successful on the day they ship, and that means predicting where’s the world going to be two years from now, which is how long it takes for anything really big to get done. Really, really big stuff seems to take that long. Like XNA took two years to really come to fruition the first time, and you had to make predictions over what that’s going to be. With Live we had to make predictions. You don’t know you’re right when you start.
-Boyd Multerer