One of the first automation challenges telecom operators must tackle is software lifecycle management, which requires mastering a dev-ops process, a leading systems integrator says.
Amol Phadke, global network strategy and consulting practice lead for Accenture, says managing software licenses and updates is a growing issue for network operators as they embrace virtualization, because their typical processes for handling updates simply don’t scale and a more automated process is needed. To date, he says, there isn’t a lot of standards work going on in the area of software lifecycle management.
“A few years ago, our clients had three or four big vendors that used to provide their equipment — routers and switches and things like that — and then provide everything that was required to run them,” Phadke says. “That ecosystem was disaggregated completely because of virtualization.”
Today, operators want to be able to embrace many more vendors providing best-in-breed software solutions that run on white boxes, but incorporating that many players into a production environment “is more than classical systems integration,” he says. Operators have to move quickly from accepting software or firmware updates on a scheduled basis, at a semi-annual pace, from a few vendors, to handling constant software updates from many more vendors.
“When [updates] happened twice a year, it was still a nightmare for the operations team,” says Phadke, who has prior experience at BT and at Alcatel-Lucent and has seen the process from both sides. “Today, you almost need constant IT, with certification labs and methodologies and testing labs and methodologies,” Phadke comments. “And that is the first area where we are doing a lot of work in automation.”
What Accenture specifically does is help their clients develop a DevOps platform; “a classical software development platform that is designed for the telco environment,” he says. “That is the foundation for all the automation capabilities operators have to build on top.”
Most current standards work is focused on automating the actual virtual network functions and enabling them to talk to lower levels of the network infrastructure, and to talk to each other in order to orchestrate a service in an automated way, he notes. But, he adds, “assuming all of that good stuff happens, we still have a massive operations and automation challenge to operate the software environment itself.”
Those efforts typically involve classic off-the-shelf DevOps tools, such as Splunk, Chef and Puppet, but it includes customization “on how to interact with the operators’ operational environment” and meet specific requirements such as ties to cloud environments, customer SLAs and more, he says.
And there is, of course, the people and culture side to the problem, with new skills required to enable ongoing support of software lifecycle management and the software production environment.