What Cloud Computing Means to Your Job

Technology has been accused of making many a job disappear, like the production line or the accounting office. And it is not done yet.

A company often resembles its communication and technology system. In the era of cloud computing that the tech industry is moving into, that seems to suggest that companies will have smaller departments, quickly analyzing data and endlessly experimenting.

That means change is on the way at the many companies that will adopt cloud computing over the next few years. Middle managers: This is not good news for you.

“Technology shapes styles of work,” said Ed Lazowska, who holds a chair in computer science and engineering at the University of Washington. “One critical advantage of the cloud is that sharing becomes dramatically easier.” He foresees more collaboration and outsourcing of work, and more specialization into whatever a worker, team or company does well.

A corporate organizational chart from 100 years ago looks like a factory, with little workers at the base like parts, assembled by managers into units that interact with or fit into larger parts. Layers of white-collar jobs died in the “corporate re-engineering” boom 20 years ago, after email and networking replaced middle managers carrying plans among departments.

Outsourcing and off-shoring happened once the dot-com bubble put lots of fiber optic cable under the ocean. Otherwise, you couldn’t have so many call centers in India, or manage a global supply chain.

In cloud computing, computer servers are pooled through management software. Power is dialed up or down depending on the workload, and the system is continually reconfigured, based on data about the next workload. To see how this changes a workplace, look at the structure of the biggest cloud companies around.

“You learn to harness feedback,” said David Campbell, the head of engineering at Microsoft Azure, the name of Microsoft’s cloud. Early on, this means lots of “A/B testing,” or putting up two versions of a website to quickly see which the customers prefer.

Azure makes engineering changes by moving parts of its customers’ traffic into the new stuff, seeing if that works as predicted and then building up. Checking expectations and hypotheses in real time, Mr. Campbell said, “takes hours, instead of months and years in the legacy world.”

Over the previous quarter, Azure delivered a new feature or service every three days, he said. “Over this quarter, it feels like it will be every two days,” Mr. Campbell said. That’s not unthinkable; two weeks ago, Amazon announced that so far in 2014, its cloud division had created 60 percent more new products than it did in all of 2013.

That challenges management. “Instead of having a debate informed by decades of experience around whether a customer would want A or B,” he said, “we define a testable hypothesis, which we quickly try to validate.” In the ego-rich halls of Microsoft, it is a lot like the adjustment to the Internet, “except the scale of everything is magnified 10 to 100 times,” said Mr. Campbell, a 20-year employee of the company.

Google also works in a data-fixated culture. Every meeting seems to be full of young engineers scrambling to amass the most compelling facts, trying to create something else they can watch customers use, then build on that. The big loser in this model may be the managers in charge of scheduling things, since it is all happening too fast.

At Amazon Web Services, which has built the world’s biggest cloud computing business, work is divided into teams of the smallest size necessary to figure out what the customer is doing with an important product. That team then quickly adapts the product to work better and looks for new insight.

“We run the entire organization as a federation of autonomous groups,” said James Hamilton, who is in charge of expanding A.W.S. “Top management can’t say ‘You must do this.’ It’s hard to give direction.” Senior executives at A.W.S. meet once a month to review what all the teams are working on, and coordinate among them.

Parts of this approach sound like the trendy “lean” tech start-up, ever changing to suit a new business model. It’s different at big companies, executives there say, when you are moving quickly but also have many customers and obligations.

“There has to be a core” that defines what the overall company stands for and does, said Renée James, the president of Intel. “You need to have more customer focus, and you need partners with a depth of local knowledge, but you have to have a strong center too.” In the last year, Intel, once the most go-it-alone of tech companies, has announced a slew of partnerships, aimed at moving faster.

It is not clear that any of the big companies has nailed this new style of work. Even so, they all say that their non-tech customers are badgering them to determine how they can break into smaller customer-focused teams, loosely collaborating and intent on moving information faster. The survivors will be good at collaboration, good at statistics and good at figuring out what the company needs to build next.

“Companies tell us, ‘We want to be Google,’” says Greg DeMichillie, the head of Google’s cloud business. “People will be more analytical, experimental, faster, more data driven at work.”