On the way to hyperscale cloud: tackling DevOps’ top challenges



Members of IDG’s Influencer Network assess the Impact of this cultural shift.

DevOps represents a fundamental change in how development and operations teams partner, as well as a set of best practices that drive automation, reliability, and speed in deploying changes. The alignment is driven by a shared understanding between agile development teams who want to release code to production frequently and by operations employees who must support high reliability, security, and scalability of applications, databases, infrastructure, and networks.

As InfoWorld has observed, automations such as CI/CD (continuous integration and continuous delivery) and IaC (infrastructure as code), platforms built using containers, and environments managed with AIops are all DevOps practices and technologies. For many organizations, maturing these primary DevOps practices may require hiring people with the needed skills, training others, testing new technologies, iterating on the implementation, and developing approaches to scale and standardize practices.

What, specifically, are the main challenges in moving to a DevOps approach and managing servers (including hardware and software provisioning) at scale?

For answers, we turned to members of the IDG Influencer Network, a community of industry analysts, IT professionals, and journalists. While their viewpoints were varied, as would be expected, Peter B. Nichol (@PeterBNichol), Chief Technology Officer at Oroca Innovations, spoke for many when he said “the top challenge of DevOps isn’t technology, it’s culture.”

“We assume that our DevOps technology will be architected, implemented, and deployed to be functioning,” he continued. “It usually is. What’s rarely discussed is the downstream organizational processes that must be changed to support DevOps.”

It’s a view that was shared by Frank Cutitta (@fcutitta), CEO and Founder of HealthTech Decisions Lab.

“One of the key questions will be whether the enterprise can operate at the speed of an effective DevOps approach,” he said. “In a stereotypical “land of slow and no” IT organization, the ability to accelerate processes at scale across disparate departments can be overwhelming from a cultural point of view.”

Larry Larmeu (@LarryLarmeu), an enterprise technology leader, agreed, noting that “the toughest part of adopting DevOps at scale is the cultural and political element.”

“Once you’re ready to scale, you have to undo years of existing practices and the siloed teams and activities that have been built around them,” he said. “Changing people’s attitudes and actions is always harder than changing technology stacks.”

David Chou (@dchou1107), a global healthcare CIO, echoed that sentiment, saying that moving to DevOps requires “an IT department change.”

“It’s effortless for leaders to say, ‘Here is a new framework to follow.’ Executing the vision and changing the culture to adapt is the hardest part. The second phase is retraining the internal staff and finding the right talent to execute DevOps.”

Automation and orchestration tools are key

A number of the Influencers pointed to the need for automation to manage servers at scale.

“Ultimately, process and tools scale, people don’t,” said Jason James (@itlinchpin), CIO for Net Health Fits. “The challenge is growing environments to scale without continually adding an army of DevOps team members. Automation and orchestration tools are needed to effectively provision and manage environments as they scale. These tools empower DevOps departments to manage thousands of servers per team member.”

Gene De Libero (@GeneDeLibero), Chief Strategy Officer at GeekHive, said “this automation of quality, security, and deployment processes across the [software development life cycle] requires an infrastructure—and ways of working—that can support current and future business needs as application usage grows. As DevOps complexity increases, organizations need software and hardware to deliver the necessary speed and scale, requiring a flexible provisioning model that includes decentralization and self-service deployment.”

At the same time, the need for automation prompted Steve Tout (@stevetout), senior advisor at Optiv Security, to urge caution.

“The biggest challenge—and risk—of automated and on-demand provisioning of servers and software is the overprivileged super admin access that comes with the territory,” Tout said. “In public cloud computing environments, the shared responsibility model shifts the burden of ultimate security to the tenants. Many of the recent data breaches have occurred because protection for highly privileged accounts is not enabled by default. To make matters worse, cybercriminals can obtain the ‘keys to the kingdom’ and use compromised and stolen credentials to walk right through the front door of restricted access systems and applications. As more servers come online, and more users log into applications running on those servers, the risk increases exponentially.”

Will Kelly (@willkelly), technical marketing manager at Anchore, similarly advised caution.

“The main challenge of moving to a DevOps approach and managing servers at scale starts with automating server management to the degree the organization truly requires,” Kelly said. “There’s going to be the temptation to want to automate way too much in the server management process. Another challenge and part of automation is moving to self-service deployment and provisioning of servers. Making self-service work further changes how cloud spending and budgets are managed”

Keeping on top of ‘all the moving parts’

For Debra Ruh (@debraruh), CEO of Ruh Global IMPACT, the main challenge in moving to a DevOps approach is “all the moving parts.” That’s why it’s essential to have a roadmap and “to get support for your teams but also give your team the budget, tools, and support they need to get this right,” she said.

“The sheer volume of all the moving parts can doom the programs,” Ruh continued. “Take the time to carefully lay out the approach, consider once again lessons learned, best practices, and work with other professionals that have already successfully completed a successful DevOps approach with all the moving parts. Making the time to allow your teams to plan this out and then operationalize so that the scale isn’t overwhelming will pay dividends in the long run. Don’t forget accessibility. Inclusion for all of your staff is essential, and accessibility also helps ensure that the processes work for everyone.”

HealthTech Decisions Lab’s Cutitta recommended a similar approach.

“Because of the hybrid nature of DevOps, the ability to share knowledge about the deployment across departments becomes critical,” he said. “This creates the need for democratization and a socialization strategy both within enterprise departments and across them.”

“Many function-oriented companies separate development and operations duties,” added Oroca Innovations’ Nichol. “Siloed functions are exactly what DevOps is attempting to collapse and simplify. Often organizations that have functional silos find the adoption of DevOps extremely difficult, especially when functional owners disagree on where the DevOps ownership boundaries begin and end. This isn’t surprising. Development and operations usually are performance evaluated based on different and near opposite measures. Development is measured on the agility to produce and release code, whereas operations are measured on platform and software stability.”

Of all the Influencers who participated in our survey, none was as terse as Scott Nelson (@Scott_Nelson19), CEO/CTO at Reuleaux Technology LLC. His top DevOps challenges?

“Agility and the unknown unknowns of the project known as digitization.”

And no Influencer was quite as vivid as CIO and CTO Wayne Sadin (@waynesadin).

“Stop thinking ‘pets’ and start thinking ‘cattle,’” he advised. “Managing deployment at scale involves defining ‘fleets’ of standard VMs and automating deployment of changes to the appropriate fleet. Of course, by the time you define fleets and associate apps with them, then automate everything, you're well on the way to hyperscale cloud, so keep going!”

Supermicro’s broad range of building-block systems are ready to power open cloud agility and the future of DevOps. Our servers are equipped with standardized Redfish APIs to automate provisioning, scale-up, and scale-out the infrastructure to support modern DevOps practices. Fully utilizing Redish APIs, our SuperCloud Composer software reduces organizations' barriers to adapt to the DevOps movement and orchestrate the open cloud infrastructure at scale. The pane-of-glass platform provides a unified dashboard that encompasses compute, storage, networking, and rack integration with rich analytics of resource pools, telemetry, and intelligent system lifecycle management capabilities. The robust node composition capabilities allow fast deployments of OS and software provisioning automation at scale with the role-based access controls ideal for modern data center security policies.

Visit us at supermicro.com/Cloud to learn more.

Organizations today are under unprecedented pressure to adapt. Intel has delivered five generations of custom silicon built for cloud scale, along with co-engineering with partners and relationships with top cloud providers. Today’s top clouds are powered by Intel® Xeon® Scalable Processors.