How To Do DevOps Without Leaving Legacy Behind

August 7, 2015 by Justin Nemmers



An increasingly apparent and large challenge in IT organizations is how teams can effectively modernize software development and IT operations while still operating and maintaining legacy infrastructure. Often the approach is to merely draw a line in the sand, creating an arbitrary cut-off whereby new implementations make use of the much desired DevOps and Agile methodology.

But what about the legacy environments?

Just because something is “legacy” doesn’t automatically mean that it’s twenty years old. Many so-called legacy systems were deployed mere months ago-- and on modern hardware, operating systems, and storage. For the sake of an agile organization, however, a legacy deployment or environment is anything that is not included in the new processes and approaches required for a DevOps-enabled organization.

The question remains: how can IT organizations successfully apply DevOps and Agile methodologies to existing legacy environments, and what are the benefits from doing this?

Start with the Infrastructure

Regardless of the type and variety of applications in an enterprise IT environment, there are likely many commonalities in the operating system and infrastructure components.

Manual OS build processes typically require significant admin-hours to deliver a single build. Additionally, the reliability of the result is a totally dependent on an admin’s experience, skill and ability to precisely follow a set of complicated directions. Then that admin needs to repeat this process over and over again for each of the systems in the environment.


There are other teams involved with build processes as well. Before the application delivery team can do their job, the information assurance team needs to validate that the correct security baseline has been applied. The more teams that need to touch a system, the longer it will take to implement, and the more likely you are to encounter delays and errors.

The system and server build process is thankfully well-understood, and easily automated. Regardless of your current OS build process (core build at provisioning, gold disk, etc.), there are likely many commonalities across your environment. Automating this build and configuration process will enable you to repeatedly deploy OS images on-demand, and, with the right tooling, manage those existing builds as easily as you create new ones so that systems will always look the same.

Once the OS build and management process have been automated, making these automations available to other teams becomes relatively trivial. Typical consumers of OS builds, such as development, testing, and QA teams, can trust they’re always working with the proper base OS configuration while building their applications.

Merely building systems, however, ignores the much harder part of the problem: keeping them updated through their lifecycles. How can you ensure that these meet the current baseline requirements as well? This is again where your choice in automation tooling makes a difference.  

A key problem with the traditional virtual machine (VM) lifecycle approach is that in the past, it has required a separate process for maintenance of existing VMs… and many provisioning tools have a difficult time updating and making changes to existing systems in a live-running environment.

Thankfully, it’s relatively straightforward to create and apply a continuous deployment methodology to the OS environment. By automating the server build and maintenance process, it’s possible to drastically decrease or totally eliminate manual steps in the server delivery process. An added benefit is also that by using Ansible as this automation tool, the same playbooks used to build systems can be used to ensure that existing systems meet the current specification.


Even before addressing the application layer, automating the creation, delivery, and management of the OS and infrastructure layer will save considerable amounts of time, and provide other material benefits to an organization’s extended teams.

Happy automating.


DevOps, IT Automation


Justin Nemmers

Justin is the former GM, Ansible, Red Hat. He has spent a career helping organizations transform their IT environments by adopting new and making better use of existing technologies. Over his career, he has held both technical, sales, marketing and product leadership roles at a number of organizations, including Red Hat, where he ran a large services team. He resides in Raleigh, NC with his wife and children and tweets from @justnems.

rss-icon  RSS Feed