Salesforce.com, Google, and Amazon say their strategies and software help enterprises move virtualized and IT resources into a cloud computing model.
Representatives of Salesforce.com, Google, and Amazon.com tried to convince an audience last night at the Cloud Connect conference in Mountain View, Calif., that the existing IT infrastructure can be moved into their cloud computing services, but a skeptical set of questioners said, "Not so fast."
"The idea behind Cloud Connect is that you can not only move specially virtualized stuff into the cloud but also IT customized stuff as well," noted David Berlind, moderator of the panel, An Evening In The Cloud, that kicked off the event yesterday evening at the Computer History Museum.
Three speakers, Adam Selipsky, VP of product management and developer relations at Amazon; Rajen Sheth, senior product manager at Google; and Peter Coffee, director of platform research at Salesforce, argued that the IT infrastructure will soon be executing in the cloud.
Amazon has the expertise to build data center operations in support of a massive e-commerce site. It can build out that capability more cheaply and reliably than enterprises can build individual data centers, said Selipsky. Customers for its raw cloud services, such as Elastic Compute Cloud, S3 storage, and SimpleDB database services now account for more demand on its compute cycles and bandwidth than e-commerce on its Web site, he said.
"Amazon cloud services are built on the same principle as Amazon.com is built," Selipsky told an audience of about 300.
Sheth said cloud users get the benefit of having additional resources when they need them, rather than having to buy them and keeping them in reserve for peak loads. "We're in the business of scale. We have to build for scale. You get to leverage that," he said, and Google cloud users gain economies of scale in the process.
"We offer not just capacity in the cloud but multiple services," said Coffee, reflecting Salesforce's new emphasis on the cloud platform as well as online application services that it offers.
One questioner, however, asked how cloud providers can deal with the inherent latency in Web operations.
Coffee said both Salesforce experts and Web speed-up experts, such as Akamai, are perfecting the art of offsetting the Web's inherent delays. Akamai caches data and content on servers that are geographically positioned to deliver responses close to their point of origin, whenever possible.
Salesforce engineers its applications to get around the chattiness of the HTTP protocol, which likes to ask a client if it's ready to receive a message from a Web server before sending it. "This isn't the old multiple round-trip Web," he said. The typical Salesforce application response time is a quarter of a second, he said.
Selipsky likewise said Amazon invests "in smart people" able to attack the problem more effectively than individual IT shops can do on their own.
"You talk about the cloud as 'your cloud,'" pointed out Tim Crawford, director of IT for the Stanford University's Graduate School of Business, one of four members of a user panel that questioned the speakers. "What if I'm unhappy with your cloud and want to move to another. How hard is it to move my data and applications?"
Selipsky responded: "We're agnostic on languages and operating systems. We're trying to do nothing that's proprietary," he said, indicating it should be possible to pick up and move to a competing cloud.
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.