In October 2005, while the rest of Microsoft was still counting the money that Windows and Office printed every quarter, a newly arrived executive sat down and wrote a memo that read more like a warning shot. His name was Ray Ozzie. He was the man who had built Lotus Notes, and Bill Gates had recruited him to be Microsoft’s Chief Technical Officer — soon its Chief Software Architect, the very title Gates himself had held.
The memo was titled “The Internet Services Disruption.” In it, Ozzie argued that software was quietly detaching from the desktop and floating up into the network — delivered as a service, paid for by the month, running in someone else’s data center. Google was already there. A scrappy bookseller called Amazon was about to be. And Microsoft, the company that owned the PC, was building its entire future on a device that might soon be beside the point.
A dog with a codename
Ideas are cheap; someone has to build the thing. Inside Microsoft, the project that would answer Ozzie’s memo got one of the great codenames in the company’s history: Red Dog. The name, depending on which engineer you ask, came from a strip club the team drove past on the way to a planning offsite — the kind of origin story Microsoft’s official PR has never been eager to confirm.
What mattered was who signed up to build it. Leading the engineering was a legend: Dave Cutler, born in 1942, the software architect who had created the VMS operating system at Digital Equipment Corporation and then, at Microsoft, fathered Windows NT — the industrial-strength kernel that still sits underneath every copy of Windows today. Cutler was in his mid-sixties and could have retired a dozen times over. Instead he threw himself at the problem of building an operating system not for a single machine, but for an entire warehouse of them — a “fabric” that could treat thousands of servers as one giant, self-healing computer.
That was the deep, unglamorous work of cloud computing. Anyone can rent out a server. The trick is making tens of thousands of them look, to a developer, like a single reliable thing — one that keeps your app running even as individual machines fail, get patched, and get replaced underneath it. Red Dog was Microsoft’s attempt to build that abstraction from the metal up.
Unveiled in Los Angeles
On October 27, 2008, at the Professional Developers Conference in Los Angeles, Ray Ozzie walked on stage and pulled the wrap off. Red Dog had a public name now: Windows Azure. Amitabh Srivastava, the executive who ran the team, posted a blog that same day titled simply “Introducing Windows Azure.” The pitch was that developers could write applications and let Microsoft’s data centers run them — scaling up when traffic spiked, scaling down when it didn’t, and billing only for what got used.
It was a striking thing for Microsoft to say out loud. For thirty years the company’s entire business model had been to sell you a box of software and let you worry about the hardware. Azure inverted that. Now Microsoft was volunteering to own the hardware, the buildings, the power bills, and the 3 a.m. pages — and to rent you a slice of it. The company that sold shovels was now, cautiously, opening a mine of its own.
Cautious is the operative word. The announcement landed with a fraction of the noise that had greeted Windows 95 or the Xbox. Azure spent 2009 in a series of community technology previews, free while developers poked at it and filed bugs. There was no confetti, no Steve Ballmer bellowing on stage. It was, in every sense, a quiet birth.
The day the meter started
Then, on February 1, 2010, the previews ended and the billing began. Windows Azure was commercially live. The meter was running. For the first time, a customer anywhere in the world could hand Microsoft a credit card and rent computing by the hour instead of buying a license by the seat.
Let’s be honest about the scoreboard, because this history series tries to be. Microsoft was late. Amazon Web Services had launched its S3 storage service and EC2 compute service back in 2006 — roughly two full years earlier — and had used that head start to define what “the cloud” even meant. When Azure went live, AWS was already the thing every startup built on. Microsoft was the incumbent playing catch-up, a role it was deeply unused to.
But being late is not the same as being wrong. What Ozzie’s memo, Cutler’s fabric, and Srivastava’s team had done was plant a seed inside a company that didn’t yet know how badly it would need one. In 2010, Azure was a rounding error next to Windows and Office. Ray Ozzie himself would leave Microsoft later that year, his warning delivered but not yet heeded at the top. The stock was flat. The press had written Microsoft off as a fading giant.
And yet, humming away in a handful of data centers, was the one asset that would eventually be worth more than everything else the company made combined. It just needed someone at the very top to believe in it — to bet the whole company on it. That someone was already inside the building, running the very division that had built this thing.
Next in the series: a new CEO takes the wheel, publishes a book, and orders Microsoft to become “cloud-first” — the Satya Nadella era begins.
Comments
Post a Comment