DevOps has emerged as the de facto cure-all for all that’s wrong with enterprise IT, breaking down silos and replacing them with agile and transparent relationships between developers and operations and seamless, fast workflows that drive greater efficiencies across IT. But legacy IT systems still need tending, even in a DevOps world. They must be secured and scaled, load-balanced, configured, backed up and monitored for compliance, without interruption to the enterprise—all considered infrastructure management tasks.
And that’s a problem, because these tasks often are handled manually using pre-DevOps processes, which breaks down the very promise of agility, speed and rapid service delivery that DevOps delivers.
Oleg Chunikhin, chief architect with EastBanc Technologies, emphatically agrees. “If an organization goes down the DevOps path, invariably old habits get carried over into the new world over. For example, if there’s a glitch with release management or a broken disaster recovery process, these problems are dealt with as they occur, or not at all, until they reach a point of critical mass.”
Chunikhin and I sat down to further discuss the issues and infrastructure management options available to any organization embarking on a DevOps strategy. Here’s our conversation.
Wolf Ruzicka: Why do infrastructure management tasks pose such a risk to DevOps success?
Chunikhin: One of the key problems is that there is a limited set of clear leaders in the DevOps tools and architectures market. There are tools for one thing (e.g. containerization) and tools for another (software configuration), and so on. But the market lacks a repeatable, standardized, off-the-shelf solution for infrastructure-level problems.
Instead, organizations could look to re-architect what’s already there. But this can turn out to be a risky, painful and expensive business. So, you’re left with few options.
As a result, making the move to DevOps becomes a challenging and risky endeavor for any CIO. If they take that risk, they end up attached at the hip to the time-consuming and manual processes that they already know—which go against the very promise of DevOps.
Ruzicka: What are the key DevOps infrastructure management considerations that IT needs to bear in mind?
Chunikhin: First and foremost, don’t delay. CIOs must recognize the problem of infrastructure management in a DevOps environment, before it’s too late. As they explore DevOps, forward-thinking organizations need to consider how its concepts and challenges can fit their organization:
- How can manual processes be eliminated?
- How can you strike a balance between the demands of the business and the need to operate in a compliant, stable and secure environment?
By addressing these challenges early on, you’ll ensure that each of these co-interdependencies is manageable. Before processes break down and critical systems are out of sync.
Ruzicka: Don’t emerging trends like cluster management and containerization address this very problem?
Chunikhin: Yes and no. Cluster management is an important part of any DevOps architecture. By abstracting out everyday management tasks, such as backups, and using orchestration to introduce efficiencies into complex environments, it’s a useful tool. But, cluster management software is notoriously difficult to set up for production use. While it can help developers manage many aspects of complex production environments, it isn’t a complete solution in and of itself.
Containerization helps with multitude of DevOps-related problems (software configuration, package conflict resolution, dependency management, etc.), but containerization alone is also not a complete solution.
Cluster management software combined with containerization, a.k.a container orchestration technologies, on the other hand, is the most recent and very promising development in the area of DevOps technologies. It promises out-of-the-box SaaS-like capabilities for fast service delivery, production quality, reliability, security, disaster recovery and so on—with less expertise needed for overarching DevOps workflows. This relieves much of the pressure on DevOps teams by eliminating the need for custom software development, lowering maintenance costs and introducing automation.
Although each solution has its benefits, neither is a panacea for DevOps infrastructure management. Today’s cluster management software, containerization and container orchestration systems differ wildly in their capabilities and are often incompatible with older/legacy software. The task of integrating each with existing back-end systems and ensuring compatibility with existing environments is a perpetual challenge in hybrid deployments, where everything must work together for seamless DevOps. In our experience, each presents unique issues that prevent them from becoming standard on any client project.
Ruzicka: Without a standard “one-size-fits-all” offering, how do organizations address their DevOps infrastructure management challenges?
Chunikhin: There are several key capabilities that organizations should be looking at:
- Portability and vendor-independence.
- Extensibility and openness, i.e. compatible with widely used modern DevOps standards.
- Easily deployable business components (such as data storage systems, deployment and release cycle automation tools, application server clusters, etc.)
- Ease-of-use and immediate readiness for production deployments.
That’s a lot to check off, but key to this approach is a combination of best practices and the use of open-source DevOps containerization and cloud technologies such as PaaS. PaaS is critical here since it uniquely provides the extensible and ready-to-use pluggable infrastructure services needed to easily configure new services, introduce development efficiencies and take care of the operational concerns that can impinge any DevOps strategy.
DevOps automation and container orchestration technologies also augment cloud strategy and multiply its benefits by an order of magnitude.
About the Author / Wolf Ruzicka
Wolf Ruzicka is chairman of EastBanc Technologies, joining the company in 2007 as CEO. During his tenure, he has overseen the organic growth of the company, taking it from 35 employees to more than 150, with staff turnover of virtually zero. He serves on numerous boards and groups, including the Microsoft Customer Advisory Board for Microsoft Azure and the World Economic Forum’s “Partnering for Cyber Resilience” initiative. Prior to joining EastBanc, Wolf served as president of APIphany (now part of Microsoft) and held leadership positions with MicroStrategy, Mercedes Benz, Lufthansa and more. Wolf holds an Master of Business Administration from Ingolstadt School of Management.