Facebook. Animoto had been using 20 servers, but quickly ramped up to 3,500 and then settled down to 1,000 as demand leveled off, he said. And rather than trying to gauge demand and buy servers to handle it, the company just pays Amazon a very affordable rate for what it uses. Beautiful.
All this was echoed by Phil Jacob of StyleFeeder, a Cambridge startup that has only taken in $3 million in funding (half from Highland Capital Partners and most of the rest from Schooner Capital in Boston). The company, which helps people with online shopping by analyzing their tastes and preferences through data mining and machine learning, says it is the largest shopping app on Facebook. And yet, said Jacob, “we own precisely zero servers.” What’s more, the firm, which uses various Amazon cloud services, handled some 20 million messages in May for a cost of about $20.
One question had to do with how cloud platform companies would differentiate themselves. By providing good performance, services, and quality of experience, in much the same way banks compete today, said Wladawsky-Berger. “I think that’s the real battleground that we should want in the IT industry,” he said.
Matthew Gray of Google emphasized that this differentiation was still at an early stage. Companies like Google have learned a lot about how to deploy applications to millions of people and make them usable. And, he noted, there’s much “value to be gained from leveraging that learning.”
Picking up on the electric utility metaphor, Sun’s Zippel stressed that “we’re still at the point of arguing AC and DC power, at that level.” What people really care about—and this point was echoed by others throughout the day—is the data, he said. “Wherever the data is may dictate where the applications run,” he said. “It’s the data that matters, not the applications.”
One of the most interesting parts of the day for me was the discussion about how disruptive cloud computing will be to existing computer companies. “Every time there’s a platform change, there’s a lot of disruption,” said Landry. Big players, he added, seldom make it through such disruptions. Landry said that unlike with some previous revolutions/transformations, big companies—in this case, Amazon, Microsoft, Google, and Sun, for example—could well emerge as winners as providers of cloud infrastructure. But he did not give them much chance of coming up with killer cloud apps. Existing firms, he said, are too locked into their current way of thinking. Startups, which don’t care about what already exists and can design with the cloud in mind, will really show the way to the new world.
