Using his excellent moderating skills, Ben championed a very interesting 45 minute discussion (mp3 recording of which can be found here) with myself, Jim Battenburg of Rackspace and John Taschek of Salesforce.com in which we each gave our views on the current state of play and a brief gaze into the crystal ball for the next 18-24 months.
Somewhat unsurprisingly, given my fellow panelists raison d’être, I elected to wear my extremely comfortable, non-itch “private cloud” cardigan and use it to represent the enterprise IT experience – both in terms of where we have been (remembering that we have been building out the private cloud here for some time now) and where I think the areas of interest and focus (no pun intended) for other large, like-minded organizations are for the rest of this year and into 2012.
During the early stages of the call, the topic of private cloud as a was put center stage with the question of “is it appropriate to use private cloud to get some of the benefits of cloud computing without going completely public”. Fair question, I thought. I know that the great private-hybrid-public debate is the source of great annoyance and or unparalleled “meh” in many circles and I know that Ben himself described private cloud as “virtualization in drag”, however, my allergic reaction kicked in pretty well when John Taschek offered up this nugget:
“our view is that there is no such thing as the private cloud because it is just an evolution of the former architecture and if you ever have to buy a box then it is not a true cloud computing platform
One might conclude that this is not a particularly unexpected response (with all sincere respect to John) from the largest SaaS provider on the planet, but what really got me confused was the inference that “private cloud = some kind of method for delivering things over port 80″ – if that’s the general feeling of how enterprises are dealing with their cloud strategies, then we really need to get some sanity back into the discussions.
When I think of cloud, I think of services. Not just applications, not even the pigeon-holed *aaS, but all kinds of services that support and enable our business. When I think of private cloud, I think of a platform that is so completely and utterly different than the one we operated in the recent past, that it is barely comparable. It is a platform that today delivers the daily requirements of our 30,000 strong user base and one that has reduced the complexity and improved our ability to respond (and also to change direction) so significantly, that I would be hard-pressed to look you in the eye and say “there is no such thing as the private cloud”.
In my response to both John and Jim’s views, I offered that my view of the private cloud was really described from that of the user in our specific context, but the more I think through it, the more simple it becomes to explain.
Take the viewpoint of the user as it relates to a public cloud service, such as AWS. When I configure my EC2 instance, attach EBS storage, build my application and then provide access to it, I have two types of user, each with a certain stake in its overall success.
1) I have the user of the AWS service (me)
2) I have users of the application that has been built (my customers)
With very limited exception (the case being that in 1) above, I know an approximate geographical location) neither of the users have any empirical information about the underlying mechanisms for which they have been provided access to the pieces they respectively needed to complete their tasks, and, as long as the performance meets their expectations, I suspect they are happy with that arrangement. Cost aside, my experience is this abstraction is often considered one of the major benefits of any promise of “cloud”.
Now, let us take the viewpoint of the user as it relates to our private cloud service (yes, that’s the one that doesn’t exist..) and strangely enough, I have two types of user, each with a certain stake in its overall success.
1) I have the user of the private cloud service (my customers)
2) I have users of the service that has been built (our business)
Hmmm. Yet again, with very limited exception (the case being that in 1) above, I know an approximate geographical location) neither of the users have any empirical information about the underlying mechanisms for which they have been provided access to the pieces they respectively needed to complete their tasks, and, as long as the performance meets their expectations, I suspect they are happy with that arrangement. Sounds almost like the public cloud, doesn’t it ?
The major difference is that our private cloud doesn’t just deal with applications. It deals with services. And, this, for the record is where the the notion of “a private cloud being virtualization in drag” leaves me cold.
Building redundant, highly virtualized data centers tied to physical locations isn’t that hard. Really it isn’t. And it ain’t a cloud.
Building a global platform for delivering services, from dynamic on-demand applications workloads through to centralized VoIP and Video capabilities that are not tied to traditional physical locations, yet meet global user expectations is hard. Really it is. And it is a cloud. A private one. It’s private because we own the hardware, the operations and we own the security. And we’re damned proud of it too.
Despite a seemingly growing misconception that the “only and correct” cloud strategy is for large enterprises to head for the public cloud and SaaS models – I’m here to fly the flag for common sense and tell you to choose the right fit for the workload, whichever denomination it may be. Yes, we use public cloud. Yes, we use SaaS. But by far the largest deployments of services we have are “internal” – this is very common for large organizations with copious amounts of legacy applications and workloads.
In a blog, Forrester analyst James Staten made 10 predictions for cloud computing in 2011. Prediction #2 stated:
You will build a private cloud, and it will fail. This is a good thing because through failure you will learn what it really takes to operate a cloud environment.
Please, please do not rule out the attraction and feasibility of building your own private cloud. Take heart and be brave. Get some pens, get a whiteboard, get drawing.
Have some fun. See if you can come up with a better one than ours (from 2008)