Companies like Melodeo, SEOmoz and startup Spiral Genetics say they wouldn’t be where they are now without access to hosted computing services.
Take Spiral Genetics. The Seattle company uses Amazon’s Elastic Compute Cloud to do genetic analysis. “We’ve seen large advances with the use of genetics for, say, cancer treatments, but it’s been limited because of the cost associated with actually analyzing the data and the time required for analyzing,” said Alindrina Mangubat, CEO of Spiral Genetics, on Monday at the TechNW conference in Seattle.
The company has developed a scalable way of doing genetic analysis using Amazon’s hosted compute services that allows it to do the analysis in a matter of hours, charging customers just for the analysis they need.
Spiral Genetics essentially wouldn’t be possible without access to compute power from a provider like Amazon, Mangubat said. “We’re a small startup that realistically just started development three months ago. What we’ve been able to accomplish is huge,” she said. “We think this kind of technology will be crucial for ushering in an age of advanced, personalized medicine.”
Melodeo, a company that lets people stream songs from their iTunes collections to a mobile phone, says it doesn’t believe it would have been recently acquired by Hewlett-Packard if not for its decision to shift to Amazon cloud services.
After helping a family member move a small publishing business to Amazon, Bob Wise, director of engineering at HP and formerly vice president of engineering at Melodeo, realized a similar move would work well for his company. “I walked back into Melodeo and said, ‘I’ve seen the future and we’re going to move,'” he said.
Everyone on his engineering staff objected, but he pushed and the company moved its service over to Amazon. The shift took several months, longer then he expected, and was “painful,” Wise said.
Still, what happened next convinced him it was the right move. Melodeo launched a marketing program with Microsoft and Bing. The program worked so well that within three hours, Melodeo had 10 times the song play that it had been supporting previously.
“Had we been running on our own hardware, we would have lost the business. There’s no doubt,” Wise said. “Then we were acquired by HP three months ago and I can assure you that if we’d had a huge customer flop instead of a success earlier this year, that would have never happened.”
SEOmoz is another company that says it might be in a different position now had it decided to build its own infrastructure. SEOmoz offers tools to Internet search marketers, including access to a link graph that shows how all sites on the Web link to one another, said Kate Matsudaira, vice president of engineering at SEOmoz. “We have 60 terabytes of data about who links to what online,” she said.
The company has been using Amazon Web Services for that offering since it first launched in beta over three years ago, she said. “It was easier than rolling it out on our own hardware,” she said. At the time the service launched SEOmoz wasn’t profitable, but it is now, and she doesn’t think the company would be if it had tried to build its own infrastructure for the service.
The companies offered just a few tips for others thinking of moving to Amazon.
Neither Razorfish nor SEOmoz run all their services on Amazon. “There are some things that we didn’t do on Amazon because it wouldn’t be cost-effective,” SEOmoz’s Matsudaira said.
A company with its own infrastructure may face challenges moving to Amazon depending on the architecture of existing infrastructure. Melodeo’s pain in moving to Amazon resulted because it was running some services on a server that was larger than the largest available at the time at Amazon. Sinking too much money and power in one place without a distributed architecture can make a move to Amazon difficult, Wise said.
Advertising company Razorfish similarly said that its internal architecture caused some issues when moving to Amazon. “We had to do a good bit of refactoring to really go from the architecture we had to taking advantage of Amazon,” said Tobias Klauder, vice president of global IT for Razorfish. Internally, Razorfish had used a tightly coupled architecture, he said. It could have moved that directly to Amazon, but it wouldn’t have allowed the company to take the best advantage of the hosted service.
None of the companies said that they’ve had trouble with Amazon handling spikes in service. Amazon has designed its service to accommodate such increases in demand, but it still would be happy to hear from customers when they expect massive spikes, said Werner Vogels, CTO of Amazon Web Services. “In terms of advance notice, there are many analogies to electricity. If you want to start up a 10-megawatt steel mill in your back yard, you probably want to give your electricity company a call. There’s a similar paradigm that we explain to customers.”
Nancy Gohring covers mobile phones and cloud computing for The IDG News Service. Follow Nancy on Twitter at @idgnancy. Nancy’s e-mail address is Nancy_Gohring@idg.com