For over five years, I have been writing about the convergence of data center, Internet and software-development technologies that has become known as cloud computing. I started writing on my personal blog in December 2006, then went on to write CNET’s The Wisdom of Clouds for the last three years.
I’ve also spent the last three years helping develop Cisco’s cloud strategy, and am just about to begin an adventure as vice president of product strategy for enterprise cloud management vendor enStratus.
Now, as I find myself honored with the opportunity to contribute regularly to GigaOM’s cloud coverage, I find myself thinking a lot about what I’ve learned in those five years. So, for my first post–and in an attempt to put some shape to my model of cloud computing–I thought I’d walk through my most-important observations to date. At worst, if I get it wrong, I hope you’ll straighten me out.
The cornerstone of everything I believe about the cloud can be summarized in one simple statement:
Cloud computing is an application-centric operations model.
What in the world does that mean? Well, let’s begin with the “cloud is an operations model” part. I wrote a post that describes this concept in detail a couple of years ago.
The operations model is one that has been discussed ad nauseum in the last couple of years, but as a quick recap, it centers on delivery of IT capabilities at scale, on demand, typically in a multi-tenant environment. It is important to understand that, while new technologies are indeed being developed for cloud, these technologies are being developed to fit the operations model, not the other way around.
The application-centric part of that statement is derived from the very nature of cloud itself. Traditionally, IT operations has been a server-centic affair:
- We buy a server
- We assign that server an IP address and wire it to a switch port
- We choose an operating system (which, I argue, is actually part of the server from an operations perspective), then install applications
- Finally, we monitor the health of the system based on–wait for it–server metrics: CPU and memory utilization, I/O rates, etc.
Now, think about consuming a public cloud service. If you don’t own the infrastructure you are consuming, you don’t own the server. You may own the operating system if you are using an infrastructure service, such as Amazon’s EC2, but for most cloud services, you won’t even have that luxury.
What you do bring to the table–er, service–is code, data, configuration metadata and/or policies that are, in fact, what makes any cloud service valuable to you as an individual or an organization. Your task in consuming a cloud service is to deliver those elements to a service that turns them into functionality that drives business value.
Thus, a new order of operations has to evolve in order to meet the demands of this new model. The diagram below, borrowed from my first post on the topic of cloud’s effects on operations, is how I see that order breaking down. Read that post to get a sense of what responsibility is assigned to each of these roles.
In coming posts, I want to dig deep into the consequences of application-centricity in cloud, and in enterprise IT, in general. There are so many interesting corollaries, exceptions and possibilities that I’m looking forward to a long conversation with you, GigaOM’s readers. Please do not hesitate to give me feedback via comments. I can also be found on Twitter at @jamesurquhart.
Image courtesy of Gary Orenstein.