Werner Vogels of Amazon on the Future of the Cloud—Quick Hits from OVP Tech Summit
There was a lot of discussion about the top trends in energy, biotech, and computing at last week’s technology summit in Seattle hosted by OVP Venture Partners. The afternoon breakout session on information technology, attended by a few dozen IT leaders, focused on the theme of “big data.” This was all about the opportunities and challenges faced by companies, research organizations, and IT departments who need to handle very large amounts of complex data, or sell software and services to do the same.
OVP managing director Mark Ashida kicked things off by talking about how managing data in the Internet cloud fits into OVP’s investment themes. “It has become much easier to link data together at a higher level, and get higher value,” Ashida said. “We’re very excited because big data and biology are converging. Lots of startups are dealing with huge amounts of data.” (By “huge amounts,” think terabytes of data per day, which is roughly the amount of information in a human genome.)
A lively panel discussion ensued, with presentations from a half-dozen tech companies from around the country that are trying to solve different parts of this problem. The list of firms—most of which are not in OVP’s investment portfolio—included Specific Media, Complete Genomics, Cloudera, Vertica Systems, and Aster Data Systems.
Then it was the godfather of cloud services’ turn to speak. Werner Vogels, Amazon’s chief technology officer, gave an overview of where the industry is headed, and what kinds of new problems will be solved. Vogels came to Amazon in 2004 after 10 years as a research scientist in the field of distributed computing at Cornell University, and has led Amazon’s considerable efforts in Web services, among other things.
Vogels first defined “cloud computing” as he sees it—a term that didn’t exist when his company launched Amazon Web Services in 2006. “There is a definition I actually like: cloud computing is a style of computing where you have massively scalable IT-related capabilities that are available as a service, over the Internet, to multiple customers.” He added that the storage and computing resources also need to be available “on demand” and on a “pay as you go” basis.
“Why would you want to go with this model?” Vogels asked. Besides being able to grow or shrink your storage and processing capabilities whenever you want to, he pointed to three main advantages of doing your computing in the cloud:
—Cost, which is most important to younger businesses and small companies. By paying only for what storage or processing you actually use, startups save money they would otherwise spend on server hardware and maintenance. (This is the most-often cited reason for using cloud resources.)
—Agility, which is important for larger companies. “If you talk to most CIOs, in any new venture they need to go into, IT is this blocker to getting things done,” Vogels said. But by using cloud resources, a U.S. company can open a new office in Warsaw, Poland, for example, without having to worry about setting up physical machines on the ground, or other IT-related headaches.
—Expertise, which is important for everyone. “Amazon guarantees scale and it will be absolutely reliable, and to move your apps into the cloud, they’ll be more secure than they were before,” Vogels said. “There’s a whole range of things where the centralized competency actually allows you a much greater comfort zone to operate in.”
Vogels then pointed to a couple of areas where he thinks cloud computing will become increasingly important. The first is for researchers in non-computing fields like biology, psychology, and history, who need to handle large, complex data sets. “You’re talking about folks who have no idea how to write programs,” Vogels said. “This is where companies like Cloudera will be really successful, in taking these really, really efficient mechanisms, and building on top of those to actually allow psychologists to analyze huge amounts of data.”
The second area, which is more familiar, is data storage and backup. Vogels emphasized that companies and organizations “need to solve reliability issues with software” because “data storage hardware will fail.” He quoted a failure rate for storage disks of 8-10 percent per year, so a medium-sized data center with, say, 40,000 disks, would need to have a staff member running around replacing about 10 disks a day.
Lastly, Vogels gave an overarching piece of advice for where cloud computing should go from here, and how it should be implemented—hinting that Amazon will continue to build new capabilities on top of its Web services platform (he didn’t give any specifics yet). “We shouldn’t forget the power of simplicity,” he said. “People say, ‘I have complex needs.’ But if the solution is scalable and rock-solid, you can make it work. Simple building blocks can be really powerful.”