Great article, Alex. I’m surprised that Mike Gualtieri thinks that cloud computing will be the undoing of the DevOps movement. I think it is quite the contrary. Maybe it’s just a different understanding of what the term means.
In my view, software-as-a-service is what started the DevOps movement. The fact that the traditional IT services can be outsourced to IaaS or PaaS providers creates a need for “infrastructure software:” software that facilitates operations such as agile deployment, quick spin-up and teardown of server instances, configuration management, and all sorts of other automation…
All these “new” operations tasks are done in code, so the marriage between Dev and Ops is natural. Another reason why the move to the cloud is fostering the DevOps movement is that with cloud services, any hiccup in operations has an immediate impact on revenue. Downtime or slow sites cause a measurable decline in customer satisfaction or, worse, customer loss. That’s why developers—particularly in the SaaS space—always code with operations in mind.
Dave Kresse is spot on with his remarks about “closing the feedback loop.” DevOps is also very much about visualization in real time of key metrics. Tools such as Librato Silverline, New Relic, Cloudkick, Loggly, etc. (disclosure: I work for Librato) can give real-time insight into server resource usage (virtual or physical), performance bottlenecks, etc. But a key difference is that it is now at the application level as opposed to the hardware level; that is the feedback loop that cloud services need.
With these tools, you can react immediately to unexpected spikes in resource consumption or a drop in performance, and within seconds you can pull up logs and time-series charts to home in on the source of the problem. I say DevOps has only just started and is here to stay.