Microsoft's Bing Outage And The NFL Parallels

Bing was down between 6:30 and 7 p.m. Pacific time due to some configuration changes during internal testing that had "unfortunate and unintended consequences," Satya Nadella, senior vice president of Microsoft's Online Services Division, explained in a late Thursday post to the Bing community blog.

Although the outage lasted just 30 minutes, it has given system administrators -- both the real kind and the armchair kind -- license to offer Microsoft advice on how to run Bing more efficiently. "I recommend pushing configuration changes to a staging server first, then to production in the future," wrote one poster to the Bing community blog. "Do you guys not have failover systems?" asked another poster.

Likewise, when an NFL quarterback goes down or a running play is stuffed at the line of scrimmage, armchair offensive linemen everywhere feel compelled to offer unsolicited advice to their televisions -- which of course can't be heard by the players -- about how the linemen could avoid a repeat occurrence. Fans don't care how tough life is in the trenches, they just want their offense to work. Same goes for online services.

As Microsoft pushes further into online services and cloud computing, it's increasingly finding itself in the position of the offensive lineman. The Sidekick outage that hit T-Mobile subscribers in October had some pundits questioning Microsoft's ability to be a cloud service provider. But in that case, while Microsoft probably did drop the ball from an administration standpoint, it was working with another company's infrastructure (Danger), and could therefore be excused for not having a comprehensive enough knowledge of its inner workings.

However, with Bing, and also with Windows Azure, Microsoft enjoys no such luxury. Azure is Microsoft's chance to show it can build a cloud computing architecture from the ground up. It's a crucial and potentially game changing investment of more than $1 billion, and it's the revenue engine for Microsoft's future. But when Azure experiences outages -- which, like all cloud services, it inevitably will -- users are going to be screaming at their screens and calling Microsoft's number.

In fact, this has already happened, albeit on a small scale. In March, Azure experienced a 22-hour outage that had some users fuming, even though Azure was a community technical preview at the time and Microsoft offered no uptime guarantees. When Microsoft flips the switch on Azure in February and begins charging customers to use the platform, outages will be met with howls of protest, but Microsoft has accounted for outage scenarios in its Azure game plan.

"We're aiming to avoid outages by writing multiple replicas of the data to multiple devices, and multiple locations, so that we have a certain level of confidence that we can limit, if not completely eliminate, the possibility of outages," said Doug Hauger, general manager of Microsoft's Cloud Infrastructure Services product management group, in an interview last month.

In the NFL, this is akin to two, or even three offensive linemen teaming up to block one particularly ferocious defensive end. One has to wonder if this strategy was in place for Bing, and if not, whether it should have been. Because each time Bing has an outage in the future, a competitor's criticism of Microsoft's cloud computing readiness gets its wings.