There has been a fair amount of talk over the past day about why companies are or aren’t choosing cloud computing. Sadly, in large part, it still comes down to confusion. Customers don’t always no what cloud computing is or why they should care. They don’t understand how the cloud is or isn’t secure, or what it really means to be locked in. Consumers — the targets of Microsoft’s new clouds commercials — they might be flat-out clueless. Part of it is just about evolution, but part of it might also be intra-community disagreement about what’s what. A unified public voice about cloud computing at a high level could be very helpful.
There was much ado on Twitter this morning about the accuracy of the cloud-provider-comparison table in our recent report on Windows Azure. You can get this gist of it, and my explanation, in the Comments to that report, but it boils down to the disagreements over definitions that have plagued cloud computing since Day One. If you ask me, the problems with attempts to tie specific characteristics to specific labels is that the constant evolution of cloud services eventually will render moot such attempts. If the main difference between Iaas and PaaS is automation, I say we paint with broad strokes and let users decide whose capabilities best match their needs.
Headlines this week were dominated by the general availability of Microsoft Windows Azure and, shockingly, even more debate over the definition of cloud computing. However, while analysts and pundits were debating how many companies will use Microsoft’s new cloud, or whether “hybrid cloud” is an accurate term, some very real work was getting underway – all under the cloud moniker. The entity responsible for it all? The United States government.
Randy Bias at Cloudscaling has written a good, some would say “provocative,” post about the accuracy of labeling connected resource pools as “hybrid clouds.” His argument is that the word “hybrid” denotes something new and holistic, not merely the connection by some means of two entirely separate systems. Technically, he’s onto something, but I think marketing needs to win out in this case. A provider or vendor trying to get the attention of C-level executives ,or journalists or whomever, needs a snappy term to get their attention and connote differentiation. Leading with an explanation of its characteristics won’t get the job done.
I must say that I am a bit shocked by the resurgence (in 2010!) of the public vs. private cloud argument, as well as the related wrangling over whether private clouds fit Nick Carr’s electric-utility analogy. In theory, private clouds should deliver the same experience to end-users as do public clouds. You write an app, port it to the cloud infrastructure, and it runs. If it’s an internal cloud, maybe you get billed by your IT department. The point is that cloud computing is a delivery model, and if users get what they need on demand, it’s a cloud to them — even if it’s just a microcosm of the electricity system. Moreover, they don’t care what it’s called.
I posted a couple of links today regarding whether virtualization is necessary for a cloud, or whether clouds can be hosted on mainframes. On the virtualization front, if users just need ad hoc resources, there is no reason a cloud could not offer physical infrastructure that is reset after each user. As for mainframes, providers need to consider what happens if one (or more) of them fails. In most clouds, a single server is inconsequential — workloads migrate and the downed box is replaced. If someone can design an equally dynamic and flexible mainframe-only cloud, then so be it. But there is no reason mainframes can’t be part of any cloud.
Two items really caught my eye this morning. The first is a post from Gartner’s Tom Bittman about coming up with a new analogy for cloud computing. He chose the transmission of music, which struck a chord (pardon the pun) because I made a similar analogy back in 2006, before “cloud computing” as a term had even caught on. It makes sense: There are benefits to both digital and physical music media, and as long as innovation continues in both areas, both areas will continue to exist.
The other item comes from IBM employee Bill Freedman. It is an insider view on how IBM is planning to spend its way to cloud dominance. I have one bone of contention with his analysis of IBM, though: IBM was not late to the cloud game, but actually was on the forefront with Blue Cloud. The question is why it failed to capitalize on that early thought leadership.
I would like to add briefly my two cents on the emerging debate over what constitutes a “private” cloud. Pundits can define and morph terms however they want, but it is customers who, in the end, will determine what means what to them. Based on my discussions with users, as well the countless articles describing their fears, private to them appears to mean “not shared.” That does not mean they won’t use hybrid clouds, just that they won’t likely be fooled into thinking they’re actually private in any real sense once the public cloud is invoked (see, e.g., my thoughts on Amazon’s VPC). Smart customers buy into the benefits of a particular solution, not the label.