DevOps is an admirable step forward for operations and development teams, but can the infrastructure guys support it? A survey from infrastructure automation software firm Qualisystems suggests that many firms have significant gaps in their support for DevOps projects.
The company surveyed 643 attendees at VMWorld in the US and Europe, ranging from technical professionals in the trenches through to C-suite staff. Seventeen per cent of the respondents said that it would take more than a month to deploy new infrastructure for development and testing staff. Another quarter said it would take over a week. That’s 43 per cent of firms that are slower than treacle when it comes to provisioning new kit.
This raises interesting questions for companies pursuing DevOps. The need for a culture that supports the process is well understood, as is the need for tools that marry development, testing and production workflows. It’s important not to forget the underlying infrastructure requirements, though, explained to Ashish Kuthiala, senior director, marketing and strategy for DevOps at Hewlett Packard Enterprise.
Automation is a key part of any DevOps project, and that must extend down to the infrastructure level, he warned.
“Set up automatic testing triggers upon code check-in, automate handoffs between teams, and carefully explore how to leverage automation to consistently deploy and configure your infrastructure,” he said. “Once something works well, codify it. Make it automated and repeatable so you can reduce errors, accelerate routine tasks, and ensure repeatability.”
Another significant outcome of the survey was the level of cloud readiness among the user base; thirty per cent of the respondents surveyed said that they were running application workloads in full private clouds today, meaning that they were not only virtualized but also had a degree of software automation for managing workloads and infrastructure.
That percentage is expected to reach 42 per cent in 24 months. Organizations appear to be rapidly adopting private cloud deployments, which could create a platform for the automation of infrastructure deployments today.
Various companies are working towards rapidly-deployable infrastructure using private cloud technologies. Hyperconvergence companies often impose a software layer on integrated components, designed for fast, automatic configuration. Firms like Qualisystems, with its Cloudshell and Testshell software products, are also targeting this problem. Firms like Cisco and HPE are doing the same with composable infrastructure. Efforts like these point to an emerging buzzword in IT: Infrastructure as Code.
This process moves CIOs and their teams away from classic configuration management, as seen in traditional physical and early virtualized infrastructures, into a new phase of infrastructure management, in which systems are accessed via APIs and provisioned programmatically.
This shift won’t be without its trade-offs. It’s probably harder than vendors make it sound (most IT usually is) and could involve changing cost infrastructures. Most importantly, it’ll probably involve new skill sets, and will definitely present new ways of working together.
Companies will have to decide, for example, how much development teams are to be involved in the provisioning process, now that all of the infrastructure suddenly speaks their language. And whoever pays the bills will have to negotiate policies on provisioning and usage of quickly-accessible resources. That’s a whole other layer of politics to contend with, before you even get to the fun stuff.
No one ever said it was going to be easy – but if you get it right, the benefits to the software development and operations process could be huge. Companies could create a continuous improvement cycle that makes software far more responsive to business needs than it can be today. But if it’s taking you a month to spin up new storage for a skunkworks application, you’ve a long way to go.