When discussing, I often think of Joni Mitchell's haunting lyrics from the song, Both Sides Now. In Mitchell's world, clouds can be wonderful "ice cream castles in the air" or annoying disturbances that "only block the sun." This duality prompts Mitchell to declare, "It's clouds illusions I recall, I really don't know clouds at all."
Joni's cloud confusion mirrors current industry bewilderment over cloud computing. Like many other industry initiatives, cloud computing has a number of meanings. Here are a few:
Cloud processing. To me, this is the foundational notion of cloud computing where businesses can rent MIPS (millions of instructions per second)-- a measure of computing capacity-- for computationally intense processing tasks. This is very attractive for basic research and should appeal to universities and small companies that can't afford supercomputers. Nevertheless, this will remain a niche market. On another note, didn't we call this grid computing a few years ago?
Cloud infrastructure. Remember Exodus and Storage Networks? Cloud infrastructure is a more modern version of these Internet boom icons. AboveNet, Rackspace, and Savvis have been making money on basic hosting services for years, but most large companies still want control of their IT assets and are willing to over-provision to maintain control. Cloud infrastructure also brings up tons of privacy concerns, just ask computing godfather Richard Stallman. This area will also remain small.
Internal clouds. The idea here is to set up an IT service and then chargeback for usage. While the cloud folks equate this to a utility services (i.e. simply plug and receive compute and storage capacity) this too is nothing new. Remember IBM's "autonomous computing" initiative? We are still a long way away from this type of simplicity.
Software as a service. Everyone points to Salesforce.com as a model of success and it truly is. Beyond CRM, there are also plenty of successful SaaS offerings for e-mail, security, payroll, etc. Ten years ago, we called these folks ASPs and MSPs. Some, like Salesforce.com, were wildly successful, but most, like Jamcracker, are either ancient history or barely hanging on. This will be where the action is. Why dedicate capital budget dollars toward on-site e-mail security appliances when Google, Symantec, and Trend Micro can provide this as an operational service?
Small businesses that lack capital or human resources are extremely likely to purchase cloud services. I certainly see this in the security market where the volume and sophistication of attacks are far too difficult for overwhelmed IT generalists lacking deep security skills. That said however, we've been throwing the utility "plug-in-the-wall" analogy around for 15 years. Yes, we've made progress toward this goal but the technology and regulatory landscape has also grown more complex in the interim. Are we any closer to this utility nirvana?
Oracle's Larry Ellison recentlyby saying that technology vendors were as fashion conscious as the women's garment industry. I couldn't agree more and have no idea why VCs are funding so many fly-by-night cloud companies. Based on my IT industry experience, IT tends to seek help in two areas: tactical high cost operations (think desktop outsourcing), or tasks that demand specialized skills (think security, Web 2.0 expertise, ITIL, etc.). Cloud computing changes the way these services are delivered but little else.