The $80 data center: cheap computing or head in the cloud?

Cloud Computing at the tipping point

Would you pay $80 (USD) for a data center with unlimited scalability and capacity? Or hold out for something a little more affordable?

Cloud Computing at the tipping point

Is pay-as-you-go cloud computing the flavor of the month, or is it a major disruption -- major disruption -- that makes it possible for organizations to tap into someone else's data center infrastructure for a few pennies, and compete head-on with large corporations that have invested millions of dollars/euros/pounds/rupees in IT?

John Evdemon calls the phenomenon "Internet Scale Computing," while Amazon Web Services refers to it as "Web-Scale Computing."

ZDNet blogging colleagues Phil Wainewright and Dana Gardner talked about the Computing Cloud concept this week. Phil explores the vision behind Amazon Web Service's Elastic Compute Cloud (EC2) and Dana breaks the news about IBM's "Blue Cloud" offerings. I think Dana's headline really says it all: "IBM's 'Blue Cloud' signals the tipping point for enterprise IT into services model."

(A little note of irony here: Amazon, which started out as a bookseller, has gotten into the IT service business, while IBM, an IT service business, has gotten into the business of selling books.)

Web services and standardization make it possible to share both IT and business services not only across the enterprise, but across the Internet. The global SOA is becoming a reality.

Privately run data centers aren't always known for efficiency. As John Evdemon puts it:

"Server farms are designed and built for maximum load conditions and dedicated to specific solutions, effectively locking horizontally-oriented resources into vertically-oriented 'solution silos.' Average data center utilization rates typically range from 15-20%. These incredibly low utilization rates detrimentally impact infrastructure and solution ROI levels."

Online podcasting service GigaVox Media, an Amazon customer, reportedly spent just over $80 in its first two months of business on storage, messaging and processing.

GigaVox uses Amazon Simple Storage Service (S3) to store files from podcasts, videocasts, and advertising images. The alternative would have been to buy disks and storage arrays to provide back-up storage, which typically range in price between $2,500 to $20,000 for network-attached storage units. GigaVox would also have had to invest in switches and hubs, as well as the expertise to put it all together and manage on an ongoing basis.

The company then adopted Amazon EC2 to fulfill its transcoding and automated show-assembly needs, and Amazon Simple Queue Service (Amazon SQS) to serve as the glue between these services monitoring EC2 server instances, queuing transcoding requests, and issuing instructions for program processing. As Doug Kaye, co-founder and CTO of GigaVox, put it in a Webcast:

"Even if we could have done this with a cluster of outsourced managed servers, which still wouldn't have been as scalable, we would have spent tens of thousands of dollars more," said Kaye. "We didn’t have to buy a single server. We didn’t have to spend any time in a ‘cage.’”

GigaVox is a great poster child for Cloud Computing, but is just that -- what about a poster adult? Cloud Computing has obvious advantages for startups and small companies, but how will it go over for larger, more established companies?

As regular readers of this blog know, I frequently talk about the "loosely coupled business," which produces services for or draws services from both inside and outside the enterprise. But some hard questions need to be addressed as we move forward:

  • Surveys I have worked with show that larger companies are not inclined at this point to tap into Cloud solutions for mission-critical applications, since the systems they have built up over the years still offer competitive advantage. In many cases, even 20-year-old mainframe programs contain custom processes and logic that provide market advantages.
  • Other than the initial rush of cost savings, will cloud computing homogenize enterprises, therefore washing away competitive advantage?
  • Any transition will still require a great deal of integration work for established companies. (Costing plenty more dollars/euros/pounds/rupees)
  • What about vendor reliability? What happens if a provider suddenly goes out of business? Or what happens if the provider decides they want to get out of IT, and contract with yet another provider for IT services?
  • And, of course, there are always worries about security, especially when files are in someone else's hands. Will Sarbanes-Oxley auditors look kindly on such off-premise arrangements?

So, a large-scale movement to Cloud Computing may take some time as these things are sorted out.

But Dana speculates that all it would take would be some large companies tapping into an enterprise service such as IBM Blue Cloud -- and seeing total IT operating costs go down by 40% over a few years -- to change the rules of the IT game forever. In addition, there will be "the productivity benefits of SOA, SaaS, services ecologies like," leading to greater agility "in how they acquire and adjust their business processes and services delivery. You might get to do more for a lot less. And a lot less IT labor."