olly - Fotolia
BOSTON -- There's talk, as DevOps takes hold, about IT operations obsolescence, but enterprise shops that use IT automation software in production say they depend on ops teams more than ever.
Large enterprises have seen an ops-led revolution brought about by IT automation, in fact, rather than ops extinction. The transition to full automation at most enterprises will be lengthy, and ops people will be the ones who actually make it happen, these companies report.
Automated systems are great until something goes wrong, said Sean Morse, software engineer at printing services company Vistaprint in Lexington, Mass.
"Developers aren't trained on operating systems and things like that," Morse said. "There are still two mindsets, and no one person can know everything."
Vistaprint has merged its dev and ops teams into one group, but the first part of this transition was to make sure developers understood their effect on operations. Then the company trained the whole team on different software frameworks so that everyone could learn new skills, Morse said.
IT automation tests infrastructure design skills
Walt Disney Studios' ops team has been far from eliminated by the company's deployment of Red Hat's container application platform OpenShift. In fact, the company's ops pros have found IT automation makes their ability to secure and network the infrastructure more crucial than before.
For example, an ops team figured out how to assign granular permissions to containers in OpenShift's Kubernetes layer by designing firewall pods that act as gateways to file storage. They also use egress pods to wall off the OpenShift system from the open internet.
Thomas HaynesLinux systems engineer, Disney
"It allows me to go back and be an infrastructure engineer again," said Thomas Haynes, Linux systems engineer for Disney, in a presentation at Red Hat Summit here this week.
With an automated Jenkins-based application delivery pipeline, Haynes said he is freed from having to repeatedly set up underlying systems for developers or deploy applications on their behalf.
IT ops still has as much work as it can handle in IT automation environments, other OpenShift users said.
"You're doing less on your own command line and taking on more of an orchestration role," said Richie McDonald, IT operations manager for a financial organization in the public sector. "There are some Unix guys who don't care for that very much, but we have no concerns about job security."
IT ops pros will be the ones to figure out how to do packet capture on inter-container networks, and add security to virtual networks such as the Open vSwitch system that underpins OpenShift, said an infrastructure architect in the financial sector, who spoke on condition of anonymity and whose company is in the early stages of deploying IT automation software.
The relative immaturity of software-defined infrastructure products, particularly in networking, will make his job both harder and more critical, the infrastructure architect said.
"If you're not prepared to talk about containers, you're going to have a bad time," the architect said. "Organizations can't afford to keep buying legacy systems because that's what you're comfortable with."
Still, "eventually someone has to plug a WAN [wide-area network] circuit into a router, and that's not going to be a developer," he said.
Products such as OpenShift depend on operations teams' skills in Linux administration, said Jamie Duncan, a cloud engineer in the public sector for Red Hat, in a presentation on OpenShift for ops at the conference.
Containers are a form of Linux that must be manipulated differently from traditional operating systems, but the same fundamental concepts and operational command line interface tools apply to troubleshoot problems, Duncan said.
OpenShift also doesn't have its own mechanism to build a new server node -- for that, IT ops must connect an elastically scalable virtual infrastructure using products such as VMware ESXi or OpenStack, he said.
"The container revolution started on a developer's laptop, but today it's being sold to ops professionals," Duncan said. "It's just another layer of density, like VMs -- you have to sweat your hardware more and automate everything, and that's the only way you're going to survive."
Selling automobiles vs. shoeing horses
Another attendee at Summit compared IT ops pros' resistance to new IT automation systems to those who wondered in the 20th century: If automobiles become safe enough for everyone to drive, who will shoe the horses?
"I would rather be a solutions architect," said Nathanael Duke, senior customer support analyst at OCLC, a global library cooperative based in Dublin, Ohio. "I would rather be selling automobiles."
Selling is exactly what IT ops pros at U.K.-based banking company Barclays had to do, said Simon Cashmore, lead engineer and solutions architect for the firm.
Barclay's platform as a service team had to view developers as consumers of the automated system ops has created rather than a foreign species of IT person, Cashmore said. Barclays has shifted people into new roles, including a "front man" to monitor chat channels and involve the right people in service requests.
"It's been a massive cultural change for my engineering team from being heads-down creating stuff to being out there working with people," he said.
And ops teams aren't the only ones who have to figure out new approaches to solving problems, Cashmore said.
Developers at Barclays have required extensive help from the ops team to teach them how to handle new infrastructure responsibilities. The company has instituted a "bring your own image" policy for developers, for example, but IT ops maintains preconfigured container images for developers to use if they struggle.
No ops job is safe without embracing change
Make life as an ops professional easier -- and better
How to work as a site reliability engineer