Can you trust your business to Google's cloud?

A large number of Google Docs users couldn't use their online word processor or presentations for about an hour Tuesday. But the glitch illustrates not just the troubles with cloud computing, but also the gradual progress in making the concept palatable.

Cloud computing, in which software runs not on PCs or company servers but instead on computers on the Internet, requires something of a leap of faith both technologically and culturally. Those making the move must get accustomed to a reliance on somebody else's computing infrastructure, and that can be scary.

What's gradually emerging, though, are guarantees and practical tools that likely will help ease the transition.

Salesforce.com shows details about service responsiveness and specifics about problems that do emerge. (Click image to see larger version.)

Google, for example, offers a service level agreement (SLA) promising that Gmail, the online e-mail component of its overall Google Apps service, will be available 99.9 percent of the time, with service credits extended to paying customers if Gmail dips below that level.

And SLAs are coming to the rest of Google Apps.

"We don't have an SLA yet for Google Calendar or Google Docs, but it's something we're moving quickly toward," said Rishi Chandra, product manager for Google Apps. Google wants "to get the same level of reliability for all of Apps," he said.

Google is a major proponent of cloud computing, with advocacy work down to the level of trying to build ubiquitous high-speed networks, and Yahoo has just formed a cloud computing group of its own. The trend has the potential to seriously redistribute wealth within the computing industry.

There are two broad categories of cloud computing. First are online applications such as Google's Apps, Yahoo's Zimbra for e-mail, Zoho for office and business software, Adobe Buzzword for word processing, and Salesforce.com for managing customer relations. Second are general-purpose foundations such as Amazon Web Services, Saleforce.com's Force.com, and Google App Engine on which customers can run their own applications.

Taking the plunge into the cloud
Service level agreements are the kind of contractual guarantees that appeal to CIOs making cost-benefit analyses. But there's a gut-level factor at play here, too.

Psychologically, it's well-known in risk analysis circles that people feel more comfortable with risk if they feel in control. Thus people are often more comfortable driving a car on a congested freeway compared with being flown somewhere in a commercial jet, regardless of the relative safety of the two forms of transport.

So naturally there's some fear with cloud computing: it means you can't reboot your laptop or check for blinking red lights on the data center servers.

Google showed this status warning during Tuesday's outage.

(Credit: CNET Networks)

Companies are working to address this side of the equation, too. One prime example is the Trust.salesforce.com site, which shows the response time for a Salesforce.com server transaction. It also details when problems happened, what they affected, and what caused them.

"We've found working with our customers they want transparency. They want to know exactly what's going on all the time," said Bruce Francis, Salesforce.com's vice president of corporate strategy. "If there's an issue, they're not furious; they just want to know exactly what's going on."

Amazon.com, too, offers a basic status report dashboard for Amazon Web Services. "A service dashboard is something our developers asked us for, and we made the service available to them as soon as possible," said spokeswoman Kay Kinton.

"Own your own risk"
And some others are even trying to make a business out of reducing the uncertainties of cloud computing. One is open-source monitoring and management software company Hyperic, which launched a CloudStatus service in June that monitors Amazon Web Services in greater detail. The company is working hard to extend its monitoring service to other sites, too, including Google App Engine, said Stacey Schneider, senior director of marketing.

"You can't get away from owning your own risk. This is slowing the adoption of the cloud," she said.

Amazon Web Services shows what's working or not. (Click image to see larger version.)

Google is trying to communicate better with users and customers, Chandra said, though he stopped short of revealing what the uptime is for Google Docs or detailing why exactly it had problems earlier this week.

"With the docs outage, we posted immediately in the administrative console that there was an issue. We posted to the help center and the phone line system that we were working quickly to resolve it," Chandra said.

Asked whether Google plans its own status dashboard, Chandra wouldn't share details but promised better help for users. "We're trying to find even more ways to be more transparent about reliability," he said.

Risks of non-cloud computing, too
Much ado can and should be made of the risks of cloud computing, but it should be noted that even the much more mature business of computing without a cloud has its risks. Downtime, either with ailing or stolen PCs or with overtaxed or faulty servers, is a serious problem there, too.

Those with high-end services boast of "five nines" of reliability, where services are available 99.999 percent of the year and therefore down no more than 5 minutes and 15 seconds per year. Google's Gmail SLA, at 99.9 percent uptime, promises downtime of less than 9 hours per year.

That might not be five nines, and it's for Gmail only today, but Google chooses to see the glass as half full.

"We talk to customers, and 99.9 percent is mostly much higher than most organizations with their internal service today," Chandra said.

[fredtheviking]

If could computing can be standardized. I believe it will be a great benefit to Business operating online. Could computing is probably less risky than managing your own hardware. Especially, if you don't have resources to manage large servers and configurations. Google has the talent the scale like few others do.

[humanssssss]

It will depend who's going to own the data and what right the provider when go out of business be forced to hand over data.

[mishmash0101]

What's the point of having a PERSONAL computer when you are 100% reliant on a server? Haven't we gone a full circle now and arrived right back at the mainframe model that we SO badly wanted to get away from? Let's just bring all of the VAX's out of retirement and say that the last 15 years were a waste of effort!

[Shankland]

You're right that cloud computing is a step back toward servers, though I'd say in some ways farther back than 15 years to in-house Vaxen to the era of time-sharing mainframes. There are some advantages to cloud computing, though: you can use a document from your PC at home, work, an airport kiosk, a hotel, or some friend's house. If your PC is lost, stolen, or disabled, you still have your documents. (The Internet is a lot more ubiquitous than whatever DECnet LAN you were plugging into in 1990, even if it's not ubiquitous enough for truly reliable cloud computing.) Cloud computing can be used, to some extent, with devices such as iPhones and Intel's MIDs that are up and coming, and they likely don't even run the same operating system as on your PC. Finally, some newer technology such as HTML5 and Google Gears let you work on documents when you're offline, so you don't necessarily need a 100% connection. It's definitely still very early days, though.

[magicmaster]

99.9% available? What's the use of online storage when it's not available?

[Tony McCune]

Interesting that we don't hear reporting about the daily or weekly brief outages at most of the fortune 1000 companies. The Amazon cloud is running at 99.9993 from the time we started using it at DigitalChalk in 2006. I'd like to see that beat in a do-it-yourself data center.
http://www.amazon.com/Success-Story-DigitalChalk-AWS/b?ie=UTF8&node=401671011&me=A36L942TSJ2AJA

[benjaminstraight]

Maybe.

[taylor_shawn]

If you ask me, the scariest part of the growing trend of cloud computing and storage is that if a big part of the system shuts down for long time, it could be catastrophic to more than just individuals, or companies, or even industries, but entire economies and whole populations. If we ALL depend on the cloud, which in time I think we will, we will all be at risk of cyber-terrorism, super-viruses, or just a simple breakdown (ok, not simple, but old fashioned I guess).

[arborlaw]

The legal risks associated with cloud computing won't go away with metrics about increased reliability. Here's what the terms and conditions for Google Docs actually say:

"You acknowledge and agree that the form and nature of the Services which Google provides may change from time to time without prior notice to you"

"You acknowledge and agree that Google may stop (permanently or temporarily) providing the Services (or any features within the Services) to you or to users generally at Google?s sole discretion, without prior notice to you."

"If Google disables access to your account, you may be prevented from accessing the Services, your account details or any files or other content which is contained in your account."

"Google shall not be liable to you for...the deletion of, corruption of, or failure to store, any content and other communications data maintained or transmitted by or through your use of the services."

This isn't the type of service level agreement that business can rely on, is it.

Carol Shepherd, Attorney
Arborlaw PLC

[lennysan]

I've done an analysis of the SLA's of the major SaaS players. If you're interested, check it out: http://www.transparentuptime.com/2008/12/saas-slas-state-of-union.html

[sbow75]

After experiencing similar data backup issues, I got together with other web developers to create a solution to this growing problem of no Web 2.0 backup plan.

Follow our progress via http://twitter.com/cloudsurance or http://cloudsurance.com