Do you need an IT Infrastructure team in a DevOps world? (Hint: the answer is yes)

DevOps is the new mantra of the application development world! If you think this is old news and you have mastered DevOps in your organization, please stop reading here. But if you are like me, and believe it's still evolving and a bit away from being the norm in enterprise environments, let's reflect on how we can enable a smooth adoption and realization of intended benefits. 

One of the questions many IT directors have asked me is this - if my developer and operations person is one and the same and I am going to consume everything in a PaaS (Platform as a Service) model, what will my IT infrastructure (Infra) team do? Today, they run the whole gamut from understanding business and application requirements from a functionality and performance perspective, designing the requisite infrastructure to support the same, procuring it in the most competitive and timely fashion, building it up and deploying it within the overall security and physical constraints of the enterprise and finally managing and enhancing it over the lifetime of the applications that run on top of it. In a DevOps / PaaS world, the concept is that the developer specifies the functional and performance requirements through a portal which automatically serves up the required platform infrastructure (be it middleware, database, server, storage or network) and does release, availability, capacity and configuration management all within itself automatically. So what will our friends in the Infra team do? Are they out of a job?

The short answer is No. As long as they are willing to evolve and align to the new enterprise platform strategy. Of course, you would always have a smaller team to support the legacy apps and associated infrastructure, in house or outsourced. For the rest, their job will change depending upon whether the organization adopts a full stack PaaS on the cloud or if they go for a DIY PaaS in-house on either a private cloud or public IaaS. You would need only a scaled down Infra team if it's a full stack public PaaS, primarily for governance (Option 1). If it's a DIY PaaS on public IaaS, the Infra team would focus on PaaS orchestration, provisioning, integration and so forth (Option 2). Lastly and quite likely the Infra team could transform into a private cloud team who will design, build, deploy and manage a private IaaS/cloud (Option 3). While options 1 and 2 are clear, how is the 3rd option any different from status quo you may ask? Herein lies the magic - suddenly the Infra team becomes more of a profit center running its own IaaS business. The “requirement-to-procure-to-deploy” cycle is literally independent of the uncertainties and vagaries of the application team. The amount of optimization, fiscal discipline and nimbleness that can be brought in is tremendous. As long as the service interface with applications can be published and allow for requisite functionalities, the infrastructure platform can innovate and improvise through adoption of virtualization, containers, cloud, on-premise utility services and autonomics.

In short, disaggregation of monolithic systems and processes and adoption of DevOps, PaaS, micro services, IaaS and autonomics will only help lean out and sharpen the Infra and App teams resulting in an agile and nimble enterprise of the future. So, yes the IT Infrastructure team is still required, but they have to change with the times.

Note: This is a personal blog. All views expressed in this article are personal and do not reflect that of my employer. This is just a perspective to provoke thought and not a fully formed argument.