It finally hit me why Infrastructure Managers are slow to adopt DevOps. They don’t see the benefit.
If your business model is based on creating and publishing applications, the benefits of the faster DevOps model are obvious. Otherwise, the benefits are more subtle and the breakeven point for making the change is farther out in time, potentially 3-5 years or more. With that said, however, traditional infrastructure managers will be glad they made the switch. The time and effort are well worth it and here’s why.
The way we access information in the future will put tremendous pressure on traditional infrastructures. Not only is the volume of data increasing, but so will the velocity at which it is created, stored, and processed. The variety of data that needs processing is also increased. The processing and presentation of that data, mostly through mobile apps, will stretch the limits of traditional infrastructure capacity.
To avoid being caught up in big infrastructure upgrades every few years, traditional Infrastructure Managers should move to a DevOps model and increase capacity incrementally, as close to real-time as possible. This is where "Infrastructure as Code" becomes important. (More on that in a future article.)
Enough with the business case… let’s get to work.
Your effort will be divided into three parts; people, process, and technology. Today’s focus is on processes, specifically, Change Management and Capacity Management.
Implementing DevOps requires these two processes to move at high speed. Here is what you need to do. First, adjust the way you manage changes. The goal is to have 95% of all changes moved into the pre-approved category. Not only will this significantly improve the speed of your organization, in the long-term it will also decrease the workload on the Change Advisory Board, the Change Coordinators, and the Change Manager.
I know, I know… this is easier said than done. After all, this is a HUGE adjustment for traditional Change Management practitioners, however, if you are ever going to move at the speed DevOps demands, you must get this done. This is not a pipe dream. I have seen it work, and work well.
Next on the list is Capacity Management. This absolutely, positively must be automated. There are some great tools available for monitoring infrastructure capacity. Deploy one of these tools as quickly as you can. The key to making this process work, however, is not in the tool. The key is creating meaningful alerts that get action. A great tool with cryptic alerts – which in turn reduce action – is a waste of time and money.
Once these two revised processes are running smoothly you can move to the next step, which we will discuss in the next post. Now… get to work on those processes and email me with questions or comments.