DevOps: The Operational Amplifier

When Instagram was sold to Facebook in 2012, it employed only 13 people and maintained over 4 billion photos shared by its 80 million registered users.

Internally, Instagram was a small business. Externally, it was a web monster. Filling the gap between those two contradictory perspectives is DevOps.

Now to be fair, Instagram (like many other web monster properties today) has it easier than most other businesses because it supported only one application. One. That's in stark contrast to large enterprises which are, by most analyst firms, said to manage not one but one hundred and even one thousand applications - at the same time. Our own data indicates an average of 312 applications per customer, many of which are certainly integrated and interacting with one another.

Which makes it difficult to manage even the most innocuous of processes. Maintenance windows exist in the enterprise, after all, to manage expectations with respect to downtime and disruption specifically because of the interdependent nature of enterprise applications.

The thing is, these numbers are only going to get worse as the Internet of Things continues to put pressure on organizations to up their app game with new ways to offer things and apps together using new business models.

Unfortunately IT budget and staff is not necessarily going to increase at the same pace. In fact, despite analysis that suggests a highly mobile customer base requires a lower ratio of IT personnel to users due to higher complexity, it is unlikely IT will suddenly grow enough to meet a ratio nearly 40 to 1 lower than optimal to support static technology users.

That means IT has to look to other means to up the output of operations teams tasked with deploying and maintaining the applications and infrastructure critical to business success.

IT needs an operational amplifier.

Operational Amplifiers
Operational amplifiers are a lot like force multipliers in that they enable a small number of people or infrastructure to achieve more, as if they were multiplied (or cloned, if you prefer).

The term comes from electrical engineering, which describes an operational amplifier as:

An operational amplifier (op-amp) is a DC-coupled high-gain electronic voltage amplifier with a differential input and, usually, a single-ended output.^[1] In this configuration, an op-amp produces an output potential (relative to circuit ground) that is typically hundreds of thousands of times larger than the potential difference between its input terminals. ^-- Wikipedia

And that is what makes it possible for a 13 member staff to support 80 million users; for a small business inside to perform like a web monster outside. That operational amplifier is devops, and it's going to be critical moving forward to shift staff from break and fix to the innovation necessary to meet the demands of the Internet of Things.

Now, that said, DevOps is not a tool. It's not a thing, it's not something tangible. It's an approach, a verb, a perspective that requires organizations to shift process burdens from people to technology in a way that makes them more efficient, repeatable and consistent.

And because there is no specific tool, but rather a mindset and methodology, it behooves producers of the infrastructure and platforms upon which applications and application services are deployed to enable operations to put into action the principles behind those methodologies: automation, orchestration and process re-engineering.

That means APIs - strong APIs - as well as extensibility and flexibility. Infrastructure cannot remain rigid and static in an environment that is rapidly changing. It must be dynamically configured, extensible, and imminently flexible.

The support of well-designed APIs and programmable data paths associated with emerging architectures like SDDC and SDN is a requirement not just for the network but for "The Network" - the whole shebang  from layer 2 to layer 7. It is through these APIs and programmatic extensibility that operational excellence is amplified and repeated across the myriad applications supported by most enterprises today.

DevOps can be the amplifier necessary to enable the economies of operational scale required to efficiently meet challenges associated with rapid, explosive growth in both user communities and app deployments. Infrastructure supportive of those efforts must provide the means by which that scale can occur.

Infrastructure must support DevOps and the shift from process reliance on people to technology through both control and data path programmability, lest it become a resistor instead of an amplifier.