THE HAGUE, Netherlands — Telcos kicked off the SDN World Congress here with boasts about how un-telco-like they’ve become, influenced by software-defined infrastructure and the world of virtualization.
Specifically, they’re starting to adopt software’s “agile” philosophy by being willing to proceed in small steps, rather than waiting for technology to be fully baked.
“The time of proofs-of-concept is over. You need to take it to production,” said Deutsche Telekom Vice President Axel Clauberg during his opening keynote here at the SDN World Congress.
That’s particularly true in network functions virtualization (NFV), where telcos might be tempted to wait for management and network orchestration (MANO) to get sorted out. Multiple proposals are in development, particularly from the Open-O and Open Source MANO (OSM) open source projects.
But waiting for a cure-all orchestrator, which would likely be the last piece in a complete NFV architecture, isn’t the way to go, Clauberg said.
“Don’t solve the biggest and most complex problems first,” he said.
The ‘Easy’ Infrastructure Cloud
Clauberg has preached SDN at every SDN World Congress since its inception in 2012, and his concept of the software-defined operator is now coming to fruition. Hence, his confidence in talking about what works.
He’s also gotten some ideas about what doesn’t work. For instance: “If you’re using a marketing slide, which looks extremely easy, you might create the wrong impression — that this is actually easy.”
That’s what happened with his idea of the infrastructure cloud, a founding principle behind NFV. He’s explained the idea for five years using the same PowerPoint slide, one that shows how network functions could be piled into “the cloud.” And it does look easy.
But as DT and other operators discovered in the years after 2012, the move to virtualization and the cloud is “extremely hard work, and it also includes a change in culture,” Clauberg said.
Specifically, telcos are full of networking experts, but what they need now are software developers. That poses an awkward transition — especially in Europe, where labor laws make it difficult to cut employees, he noted.
But he added that the goal shouldn’t be to simply drop experienced employees; rather, they need to be retrained for the software age. “We have an ethical responsibility toward our employees. We cannot just swap out our employees.”
DT’s goal in adopting NFV and software-defined networking (SDN) is to deliver new telecom services and to do so more quickly. That’s going to require a transformation of the data center, a goal that DT and other carriers share with the hyperscale web players.
To that effect, DT is interested in the Open Compute Project. Clauberg encouraged his fellow service providers to get involved as well, as he thinks the model and the approach — as well as the use of open source components — points toward the future of the data center.
Along similar lines, he gave a pitch for the Telecom Infrastructure Project, which was formed at Mobile World Congress in February and now has 190 companies as members. That group’s first summit will happen Nov. 1 in Menlo Park, California, at Facebook’s headquarters.
KPN’s Small Steps
Clauberg was followed by André Beijen, head of network innovation for KPN, the Dutch incumbent telco.
KPN likewise had to adjust its processes for the software world, and Beijen agreed that the key was to embrace the trial-and-error cycle, rather than waiting for a complete, perfected architecture.
“That’s new for telcos. Normally we plan these huge changes,” he said. “This will not work for software-defined networking and network functions virtualization.”
To handle the transition to a software-driven world, KPN took the approach of asking for volunteers rather than simply assigning people to virtualization projects.
“That worked quite well, because you get the people who are really committed to this new change,” Breton said. “You don’t get [only] the people who are available. You get the people who want to be there.”
He also stressed the importance of pilot projects. For KPN, they’ve been a way not only to test out new technology, but also to test the people — to learn from the way that new processes were (or weren’t) accepted and implemented.
Another key to the whole process was a willingness to “throw away some of the innovations you’re building right now” if they turn out to be dead ends, he said.