Colt Technology Services cues up several big projects in early 2019

Colt Technology Services will kick off 2019 with several key initiatives that it has been building toward over the past few years.

Mirko Voltolini, Colt Technology Services

The launch of Colt's universal customer premise service (uCPE) has been pushed back from late this year into the first quarter of next year. The uCPE service is the first phase of Colt's cloud-based, network functions virtualization (NFV) Stratus project.

In June, Adva announced that Colt would be using its Ensemble Connector as the network operating system for uCPE and third-party applications and services, including new virtual network functions (VNFs), SD-WAN and new virtual firewalls.

The Stratus project, which includes uCPE, is the culmination of Colt's work on NFV that dates back several years. Colt has taken a deliberate approach to how it is deploying and using both NFV and software-defined network across its on demand platform.

With uCPE close to deployment, Colt is also working within Stratus on network functions infrastructure (NFVi) to put compute capabilities at the cloud edge and then NFV orchestration over the top of all of the various functions and services.

The goal of Stratus, which was started earlier this year, is to move beyond basic connectivity by enabling value-added virtual network services (VNFs) that can be spun up or down in real-time.

In addition to Stratus, Colt's Novitas project, which started more than three years ago, handles the SDN orchestration, including automated real time service quoting, ordering and delivery, service capability development, and SDN federation.

"Deploying a proper, OpenStack-based telco cloud will allows us to scale up or scale out," said Colt's Mirko Voltolini, head of network on demand, in an interview with FierceTelecom. "In addition to that, we also have a universal CPE, a component which is again, a similar concept but is a micro-cloud on customer prem. It will allow us to host the multiple different types of VNFs. So we already have a telco cloud, but Stratus is the next generation telco cloud for us. It is iteration number two.

"Stratus will be linked with Novatis in the sense that Novatis now is the overall orchestrator of orchestrators. Stratus will kind of fall into Novatis, in the sense that you will be able to expose to customers this on demand, virtual functions through the Novatis portal and orchestrator."

Colt is targeting the first quarter of next year for the next phase of Stratus' NFVi deployment. Voltolini said Colt's NFVi was loosely based on ETSI's definition, but it has developed some of it internally.

In addition to the uCPE, Colt is looking at launching three Ethernet services in that same time frame as well as a Layer 3 IP enabled service on its on demand platform.

Novitas features a portal and API layer, the latter of which uses the MEF's LSO (Lifecycle Service Orchestration) API for connections with third-party networks. Colt is targeting the first quarter of next year for inter-carrier orchestration Ethernet services with carriers such as AT&T and Verizon.

RELATED: Colt tees up intercarrier Ethernet services with multiple service providers

In June, Colt announced the launch of its U.S. network in 13 cities across 30 data centers. Colt expects to launch its on demand bandwidth service in the U.S. in the first quarter of 2019 as well.

"Today you can buy on demand connectivity from Colt with this Novatis capability," Voltolini explained. "You buy data center to a retail building for an Ethernet service with maybe 10 gig connectivity. When we launch this Stratus capability, you can add a VNF on top of it, and make it operate a as a service, such as an internet access service, on Layer 3 with virtual firewalls. So you can then pick and choose different components, service components, like a Chinese menu. So the two things (Novitas and Stratus) will be linked to each other."

Colt launched the SDN-based Novitas in September of 2015 and then went live with it across 30 of its own data centers in January of the following year with an initial use case of an Ethernet service from a data center to another data center. 

"We now cover over 20,000 buildings and about 600 data centers," Voltolini said. "We expanded too in terms of countries, so we went from Europe to Asia, to Japan, and to Singapore and next is the United States."

Colt rolled out its first on demand service last year in Europe and then in Japan this year. Colt’s on demand offering is an enterprise-grade service that has generated a lot of interest from Colt's enterprise customer base, but it is also seeing interest from wholesale carrier customers.

Colt announced in August that its on demand service was available in Singapore while also launching SD-WAN in the Asia-Pacific region and the U.S.

The third leg of Colt's on demand platform is called Sentio. Sentio is tapping into artificial intelligence and machine learning to make the network more self-healing and more self-sufficient en route to a fully autonomous network down the road. 

"Sentio is trying to introduce automation in the service assurance space," Voltolini said

"How do you control automatically the service without manual intervention if something doesn't work as expected or if you need to fine tune the performance of the service? And so the intent is to introduce some machine learning, AI based automation into the way that we handle this service assurance, into the management and change control processes, which can take multiple different phases."

Colt has already conducted a proof-of-concept for Sentio with Ciena, and it has started to delve into two uses cases over the past few months.

"They (Ciena) have a tool called Network Health Predictor, and they've been adapting that tool to our specific needs," Voltolini said. "It's also early stage for them so we also have the opportunity to work with their R&D to shape what they have to our needs.

"We are working with a few other companies I won't name yet, because we are still in commercial discussion to build a generic data lake and a generic software layer, which will allow us to build a different algorithm models that will allow us to set up specific actions and use cases."