CenturyLink’s network virtualization and edge compute strategy has been the talk of the town over the past week, in the process taunting AT&T’s own virtualization project. CenturyLink could be left red-faced if all this trumpet blowing proves to be purely hot air, particularly with the US telco’s inferior capex and recent gaffe which saw its network crippled for a day and a half.
In short, CenturyLink has just switched on over 100 initial edge compute locations around the country for managed services, including a 5ms transport time from existing locations to the edge which it highlights as ideal for low latency use cases including VR, AR and machine learning-based services.
But then a little spice was added to the mix. Speaking at a Cowen and Company event, CenturyLink CTO Andrew Dugan was visibly miffed about AT&T’s strategy and provided some insight into the comparative strategies. “I’d like to figure out what AT&T means by 75% virtualization. I don’t get it. The concept of virtualizing the core router or an optical platform, that’s a lot of cost of your network to provide services. We’re not working on virtualizing that stuff,” said Dugan.
Compare this with the less mercurial sentiments of CenturyLink’s senior director of strategy and advancement, Bill Walker, towards the end of last year. “We don’t convert the legacy until we get our money out of it. So, unless we see a viable need or business case to replace how we manage a core network, it’s not going to be virtualized quickly,” he said.
Dugan is apprehensive on putting a number on the percentage of the network CenturyLink is preparing to virtualize, despite it being substantially smaller than AT&T’s. He explained his rationale by saying, “When you build out an NFV platform, you have the cost of the white box, you have the cost of the management or virtualization software that runs within the white box, and you have the cost of the virtual functions themselves. If you’re running one or two applications on premise, it’s not cheaper. The real value from NFV comes in the flexibility that it provides you to be able to put a box out there and be able to turn up and turn down services. It’s not a capex reduction. It’s a reduction in operating costs because you’re not having to roll trucks and put boxes out.”
There have been no recent announcements regarding network vendor supplier relationships, although Adtran we know has supplied CenturyLink with fixed network equipment in the past. While Indian network outsourcing group Velankani told us at Broadband World Forum last year that it has been successful selling CenturyLink products covering service orchestration which flows through order management, resources management, billing, and service provisioning. Velankani has come up with a neat idea of mapping broadband operator capex into the real world with its NOCPlan which is being used at another US telco, Windstream Communications. Versa Networks is another key supplier, delivering its fully managed software-defined wide area network (SD-WAN) technology, while Infinera made the nodes required to switch packets over fiber optic services across CenturyLink’s six long-haul networks comprising its US backbone.
CenturyLink, as of last year, also became an active contributor to the open source community by donating its NFVi orchestration process, having extracted the Service Logic Interpreter from a module of the ONAP (Open Network Automation Protocol) platform which we covered last week.
Of course, this comes full circle to enabling video streaming services. At the time, in April last year, Jack Pugaczewski, distinguished architect at CeturyLink said, “So if we want to move to a Netflix or Amazon sort of customer experience where the customer spends 30 to 90 seconds activating a service, just like you would to get a movie, we have to have automation. These APIs and the corresponding software that’s being built is going to give our customers that 30 to 90 second Netflix or Amazon experience instead of 30 to 90 days.”
As noted in our opening gambit, a postmortem of CenturyLink’s catastrophic network outage back in December, released this week by the FCC, has revealed that just four bad packets were at fault for a full meltdown of its broadband internet and VoIP services over 37 hours. CenturyLink appears to have avoided any fines but is in the process of updating its nodes to reduce the chance of the transmission of a malformed packet in the future.
CenturyLink came out with second quarter results last week, showing broadband subscribers subsiding slightly by 56,000 from the previous quarter to total 4.75 million, representing a 156,000 year on year decline. CenturyLink’s video business has also been on the ropes, pulling the plug on its CenturyLink stream OTT video service in March 2018.