"Amazon, at its heart, at its core, is a technology company," said Werner Vogels, the chief technology officer for the e-commerce giant, during an open house event last night at the company's new South Lake Union headquarters in Seattle (where Xconomy is putting on a feature event called VC Crossfire on October 28). "We joke that we are a technology company, and that we happen to do retail."
Essentially, founder Jeff Bezos was fascinated in the early days by how the Internet could make it possible to do things in retailing that couldn't be done before, Vogels said. A great brick-and-mortar bookstore might have 25,000 or 40,000 books, which seems like a lot, but it's nothing compared to how many books have been published throughout history. Lots of people like to read, and they like the latest books, but many people also like to dig up obscure titles that you can't find at even stores with the best selection. "It was a unique proposition, and we needed technology to make that work," Vogels said.
In the early days of Amazon's technology history, say 1995 through 2001, "Amazon in essence was a bunch of app servers and databases," Vogels said. Getting big fast and grabbing market share was more important than "architectural coherence," Vogels said. There were databases held together with "duct tape and WD40 engineering," he says.
By the 2000 and 2001 holiday shopping seasons, when Amazon saw big surges in customer demand, the company decided things had to change. "We realized we couldn't scale any further" on the existing architecture, Vogels said.
The company's guiding principle was to focus on serving the customer. On the technology side, that meant having infrastructure that was strong enough to scale up to meet huge ups and downs in demand, cheap enough to keep prices low, and reliable enough so that if a tornado hit one of the company's data centers, customer data wouldn't be lost. By 2004 and 2005, this vision for "cloud computing" started taking shape at what is now called Amazon Web Services.
Having secure, remote servers to handle huge peaks in demand during the Christmas season meant that Amazon had a lot of excess capacity during other times of the year. And that's what has enabled the company to rent out its cloud computing capacity to all sorts of small tech startups, helping to drastically lower start-up costs for hundreds of Internet companies in recent years. The Amazon cloud computing infrastructure even caught my eye (as Xconomy's biotech editor) this summer, when I started hearing that more and more biologists are using this service to store and access vast piles of genomic data coming off the new breed of fast and cheap gene sequencing machines.
A key to juggling all these tasks was to make as much of this cloud service automated as possible, Vogels said. During holiday seasons of the past, when Amazon engineers were asked if they could return some of the excess server capacity after Christmas, they often said something like "No, we'll hang onto it, just in case we need it," Vogels said. But it was inefficient to sit on all that capacity, and better to design a system that could automatically scale up and down.
"There's no engineer anymore involved, there's no emotion, just business rules decide how to use the capacity," Vogels said.
Vogels' talk was aimed largely at the engineers in the open-house crowd, and got fairly technical at times. But as a business writer listening to Vogels, it does strike me as a trend of lasting importance when companies like Amazon and Microsoft make big investments in cloud computing infrastructure. It's the sort of thing that will make it possible for startups to rent processing and storage capacity and pay as they go, instead of spending a lot of precious start-up capital on their own equipment, freeing up more time and money for whatever the real business idea is.
And Vogels stressed that Amazon's same customer-focused mentality in the retail business extends to how it treats business customers who use its cloud infrastructure. "In Web services, it's about security, reliability, scale, and low cost," Vogels said. "These are things that our customers will forever appreciate. So you put a lot of energy into innovation, that continues in those dimensions."
If Amazon is successful at doing that, economies of scale will improve, and prices will come down even further, just as with Internet retail. "We're applying the same principles," Vogels said.
More from Xconomy.com:
- Startup School: Round Two: The 2010 Xconomy Guide to Venture Incubators
- Why Facebook Places Will Make Foursquare Into a Footnote
- Kendall Square Wants an Entrepreneurial Walk of Fame and So Should Every Innovation Hub
Luke Timmerman is the National Biotech Editor of Xconomy, and the Editor of Xconomy Seattle. You can email him at firstname.lastname@example.org, or follow him at twitter.com/ldtimmerman.