The Ongoing Softwarization of Telecom
Thoughts from MPLS+SDN+NFV Paris March 2017
Kudos to the organizers who do a great job of keeping this long running event fresh and relevant, even as they continue adding acronyms to its name. I had the distinct pleasure this year of presenting an opportunity for SPs to offer cyber security as a managed service to business customers, by exploiting network function virtualization.
My major takeaways in the form of five “headlines”, are:
Wiktionary defines softwarization as, “the use of a software solution, rather than traditional hardware, to solve a problem.” With that definition in hand, the softwarization of our industry is an fait accompli (to use an appropriate French term) at least from the perspective of this conference. The entire first day of plenary presentations focused on the role of software – under the ever broadening umbrellas of SDN and NFV – for controlling and managing networks, and creating and delivering services. The underlying network was simply assumed. If you want to be a thought leader, talk software and the value you derive from that.
All Hail Open Source
Everyone publicly claims to support Open Source as a key element of the evolving software framework, even though private conversations revealed that many with closed software interests are biting their tongue as they say that. It has become religious. It was claimed that we can now build an entire modular mobile solution on top of an Open Source-based CORD architecture encompassing SD-RAN, ubiquitous EPC, and even MEC. The challenge is how to do this effectively. To paraphrase a speaker, "All the pieces are in place, we just now need to figure out how to put them together." Areas for future work are around open source business and pricing models, and how hardware-based companies can transition their business to software.
SD-WAN is the Hot Application Du Jour
It has replaced or subsumed vCPE. Stated another way, these two technologies-solutions are merging, where it is now assumed that a vCPE solution will support SD-WAN, or vice versa. All the major SD-WAN players were present, and as expected the definition of what is important in an SD-WAN feature set varies on whom you talk to. In this author’s opinion the core offering is application-level traffic management and associated cost optimization over a dual MPLS-Broadband network. Based on this definition it is still early days for SD-WAN. A presenter’s slide showed that that the total SD-WAN managed broadband traffic is still miniscule when compared with total MPLS traffic. But looking at this another way, while SD-WAN’s business impact to date is still small, it has a huge base to cut into.
Autonomous Networks may One Day be Real
Autonomous networks are creeping up the hype curve from pure vision, to something that we may be able to start implementing in niches of the network in a few years. Technologies like analytics and artificial intelligence are key, and need to continue being advanced. In an AI world an autonomous network may make optimization decisions that a human would never consider. Juniper threw out a challenge for someone to construct an autonomous data center that runs for six months without any human intervention as a place to start.
Networking Hardware still very much Matters
While thought leadership focused on software, no one was downplaying the role of routing, switching, and transport hardware. These fundamentally underpin networks, and will continue to generate the vast majority of supplier revenues. Many exhibitors displayed and talked up the features of their boxes, and a conference track was dedicated to MPLS. According to my colleague who participated in this track many presentations discussed the conceptual shift from treating the label as a transport identifier to using it as an instruction set, thus bringing “programmability” into the network. Other areas of focus were on segment routing, and telemetry for reporting network states.
Click here to see the video of EANTC's Interoperability Showcase from MPLS + SDN + NFV World Congress.