'Cloud' computing's reliability gap

Online software may be the future of computing – but the truth is, it's far from perfect.

• February 12: Research in Motion's BlackBerry e-mail service goes on the blink for three hours, and slows again a week later.
• February 15: Problems with Amazon Web Services' S3 online storage service takes several sites down for two hours.
• February 24: Google's YouTube video service is knocked offline.
• February 26: Some customers of Microsoft's Hotmail e-mail have their service unavailable for several hours.

In January 2006, Salesforce.com (CRM) faced a crisis. Service outages at the online software company had rankled customers, and threatened to spook investors and send the stock skidding.

For a business whose model relies on companies that ditch desktop software for Internet-based services, this was a serious problem. Unless Salesforce could assure the public that its online business management tools were reliable, it risked losing its momentum and giving the entire online software revolution a bum rap.

Fast forward two years, and recent headlines suggests the reliability issue hasn't gone away. Several high-profile blackouts hit the Internet's most popular services in February, including YouTube (GOOG), Hotmail (MSFT), Amazon (AMZN) and BlackBerry (RIMM) e-mail. But while popular services are struggling to keep the lights on 100 percent of the time, companies are also discovering that honesty matters more than perfection.

Salesforce, for example, learned that lesson after its January 2006 system failures. Executives realized that a lack of information during the outage had frustrated customers as much as the outage itself, and decided to take a bold step: The company would let the public see exactly how its system is functioning every day, sharing embarrassing details every time the system slowed or stopped working.

"This was opening up the kimono to our customers in a way that really hadn't been done before," said Bruce Francis, vice president of corporate strategy at Salesforce.com. "But what's more frustrating: the fact that there's a problem, or that you're not getting the right information? Usually it's that you're not getting the information. When we realized that customers just wanted that transparency, we gave it to them, and that really helped build trust."

There was more at stake than just goodwill. In recent years, online services have emerged as an important profit engine for the technology industry, as tasks from payroll to video editing move online. Large players such as Google and Microsoft are placing billion-dollar bets that in the future, many more people will store their data and run their programs through "in the cloud" through the Internet, rather than solely through desktop software. But for that vision to become reality, customers will have to remain confident that Internet-based systems will work when needed.

"Culturally we've gone through a rite of passage where we now trust that in-the-cloud services are going to be there at all times," said Bill McNee, CEO of Saugatuck Technology, an analyst firm. And despite the glitches, McNee said, that faith is well founded: "The level of reliability and availability is substantially better with these online services than it is within most companies that try to run their own systems."

And it's getting better all the time. Zach Nelson, CEO of online business software provider Netsuite (N), said that while the best service providers are always striving for perfect performance, they also learn a lot when things go wrong.

"These failures in services like Hotmail and YouTube and others? Guess what, it helps you learn about the process of delivering it more efficiently next time," Nelson said. "We've come a long way in a very short time."

Still, customers have their limits. Before Amazon's online services faced problems earlier this month, the company had already realized it needed to create a site, similar to Salesforce.com's, that displays the health of its system. While the interruption was short, and the company said customers mostly took it in stride, customers also made their displeasure clear on message boards. "My business is effectively closed right now because Amazon did something wrong," one person wrote during the outage. "I'll have to reconsider using the service now." Companies may learn from failure - but so do consumers, and not always the same lessons.