There were two conferences in the San Francisco Bay Area last week with content targeted at cloud-computing consumers. These two conferences, Cloud Expo and QCon, helped me to articulate a trend I've been noticing for some time; the cloud market may be sending very different messages to IT operations audiences than it is to software developers.
I attended Cloud Expo (while I simply tracked QCon through Twitter), and I agree with Jay Fry that this conference has gotten significantly better than its early days. It is important to note, however, that the content was most often geared to IT operations professionals, chief information officers and chief technology officers.
What was striking to me last week was how many vendors were pitching "here's how to replicate in the cloud what you do in your existing data center environment today". The pitches generally relied on terminology that most existing IT professionals are comfortable with; things like "CPU utilization" or "WAAN optimization" or "VM management."
Don't get me wrong, there were exceptions. Some vendors have discovered that their technologies can bridge the gap from infrastructure operations to service or even application operations. So they were positioning their products as useful in strengthening a cloud service offering, or providing a valuable service to an application system. There were also some professional services companies that clearly understood how cloud changes software development and deployment.
But I was disappointed again and again with how few established vendors have left their server-centric past and embraced the application (and cloud service) centric world of enterprise cloud computing.
On the other hand, QCon was generally developer-focused (covering much more than just cloud). The cloud-related presentations I saw via Twitter were generally focused on developers building for public cloud services. Sessions like "Netflix's Transition to High-Availability Storage" and tracks on NoSQL and "Architectures You've Always Wondered About" contained sessions that deconstructed the way we've always designed, built, deployed, and operated applications. Cloud was often a major influence on new, disruptive approaches or technologies.
Where Cloud was directly discussed at QCon, the conversation was generally about how to use new technologies and techniques to not only build your cloud applications, but to deploy and operate them as well. There were few sessions that discuss how to replicate existing processes and policies in the cloud. There were many about how to rethink core concepts for an entirely new scale of operations, and much more agile access to computing and storage resources.
Here's the crux of my argument. Developers are leading the charge to cloud, whether IT operations likes it or not. Cloud computing is an application-centric operations model, and as such its adoption is driven by how applications are built, packaged, deployed, monitored, and automated. Re-creating the server-, network-, and storage-centric approaches of the static past in a cloud environment is not conducive to meeting the demands of this new operations model.
Yes, the disruption caused by cloud means we are often ignoring hard-learned lessons of the past when we "simplify" operations to make application development "easier." Yes, we are breaking core assumptions behind existing security, management, and development "stacks." No, we shouldn't throw away all of the great technology that exists to support the infrastructure and services layers of the cloud operations stack.
However, those vendors with tools to market to cloud providers (public and/or private) or cloud consumers had better begin understanding how "application-centricity" affects their target markets, product messages, and even product roadmaps. "Re-creating" an enterprise data center in the cloud is not the ultimate destination here. Rebuilding our IT models to adjust to and benefit from a new, disruptive but highly valuable operations model is.