virtualized x86-based computers so that users could do things like run Linux on their Windows machines, and vice-versa.
This was no mean feat, given that Intel had never really designed x86 chips to be virtualized, and that the PCs of the time had to manage so many different peripheral devices. Eleven years later, the achievement would bring Rosenblum and VMware’s other technical co-founders the prestigious ACM Software System Award.
Workstation depended on having a host operating system such as Linux or Windows present to handle input/output operations and other actions. But at the same time, the VMware founders also built a bare-metal version called ESX—for Elastic Sky X—that managed I/O directly. This was VMware’s first “hypervisor”—a virtual machine manager that runs directly on hardware, without a host operating system. And it was the key to opening up a huge new market for the company: corporate data centers.
First released in 2001, ESX quickly became the core of VMware’s enterprise virtualization business. (Today its successor, ESXi, is part of vSphere.) Demand for the technology was epic, allowing the company to grow to 1,500 employees by 2005. Greene, the company’s founding CEO, “was able to gather an entire industry around her vision, based on her husband’s research,” says Jacques, who joined in 2006 and is something of a company historian.
Another key legacy of Greene’s leadership was the independence she won for VMware even after storage giant EMC bought the company in 2004 for $625 million. Says Jacques, “She insisted that EMC not treat us as a subsidiary but allow us to continue as our own entity”—an arrangement she believed would bring VMware the heft of a large company without sacrificing the nimbleness of a startup.
Less than three years after the acquisition, Greene engineered an unusual public offering in which the company put 10 percent of its common stock up for sale on the New York Stock Exchange, raising almost a billion dollars. The stock soared—briefly giving VMware a market capitalization approaching that of its parent, EMC.
It was a remarkable first decade. “The entire company would have stepped in front of a bullet” for Greene, Jacques says.
Alas, that wasn’t enough to offset rising antipathy between Greene and EMC’s CEO, Joe Tucci. According to a New York Times report at the time, Greene “believed Mr. Tucci did not understand VMware’s business or her contributions, and he viewed her as too demanding.” The two were also reportedly at loggerheads over Greene’s desire to spin out VMware completely.
At the same time, VMware had begun to miss earnings targets; its stock price had plunged; and some 30 to 40 percent of corporate data centers were already virtualized, leading people inside and outside the company to ask how much room was left for growth. Says Jacques: “There was a big question among customers, investors, and employees around: are we a one-trick pony? Is there more to VMware than just virtualization?”
Greene was let go in July 2008. Paul Maritz, a former top executive at Microsoft whose personal information management startup, Pi, had just been acquired by EMC, was appointed to replace her.
The key question, as Maritz stepped up to the plate, was what the company should do next. Maritz approached the question as if he were a venture investor, Jacques says. “He went out and said, ‘I can see some trends in the industry. I am going to make a range of bets, some of which will pay off, some of which won’t.’ He brought in some very experienced execs to gain a level of operational discipline that we didn’t have before. And at the same time he made a series of acquisitions that in many ways gave us an option on the state of the world.”
Key acquisitions included SpringSource, a provider of tools for enterprise Java software development; GemStone, a specialist in managing distributed data; DynamicOps, which makes tools for automating private computing clouds; and Nicira, a leader in network virtualization software.
Many of Maritz’s moves were designed to position VMware to respond to the rise of public cloud computing. Amazon had introduced its Elastic Compute Cloud, which allows users to rent virtual machines on Amazon’s infrastructure, and S3, its online file storage service, in 2006. San Antonio, TX-based Rackspace began offering similar services the same year. In this new era, any developer with a credit card could line up processing and storage capacity on demand.
And many were doing just that. Freed of the need to buy their own hardware, thousands of startups were able to experiment with new types of consumer- and business-facing Web and mobile applications. That initiated a wave of innovation that still hasn’t crested—and eventually, bigger companies started to get jealous. They’d been investing in data center virtualization for years, but still found it too complex and difficult to assemble internal computing resources quickly, at the moment they were needed.
Some of those companies began to stray off the reservation. “There have been large-scale customers who have IT people who have decided to branch out and use other services,” says Mike Adams, a group manager at VMware who handles product marketing for the vSphere platform. “They want to see what the public cloud has to offer.”
That posed an obvious threat to VMware’s business, which is built around license revenues for its on-premises software, and the service contracts that go with it. (Last year, 92 percent of the company’s revenue came either from licensing or software maintenance.)
Virtualizing Everything
2012 was another year of transition for VMware, as Maritz handed the reins to Pat Gelsinger. The former Intel executive, who joined EMC in 2009 as chief operating officer for information infrastructure products, is known in the industry for his ability to move organizations from the product brainstorming phase to execution. So if one thrust of the Maritz years was to let a thousand flowers bloom, Gelsinger’s job would be to pick just a few of them.
It was clear that one major theme for VMware under Gelsinger’s leadership would be automating data centers at a level beyond that afforded by
