Delivering application performance and making network connectivity match the business customer's specific needs are among the priorities for Global Capacity's new CTO, Tony Thakur. The telecom industry veteran, with stints at tw telecom, CENX and Telepacific under his belt, says his new company is in the process of a major network upgrade to enable it to do more than provide the connectivity for which Global Capacity and its One Marketplace approach have become known. (See Global Capacity Names New CTO.)
"In addition to providing the bandwidth and the network connectivity, cloud providers want to know there is no downtime, that everything is seven by 24, and you need to be able to manage and monitor their applications across the network," he tells Light Reading in an interview. "I think that is going to be a critical item going forward for application providers."
That includes delivering low latency to applications that depend on distributed compute and storage, he adds, citing examples such as healthcare applications and financial services, which depend on fast reaction times. And it means having the network management in place to ensure applications are given the treatment they require.
Global Capacity already has a network capable of delivering large bandwidth applications, Thakur says, so it is now working on transforming that network through several projects designed to support the application-specific support that will be required.
"We have all the raw materials, I think," he says. "Fundamentally we're creating an Ethernet backbone that has the capability to scale to deliver these applications that require low latency and more. We are upgrading our MPLS core and our latest routers and that's to create a few things: One is for creating some products, Layer2-, Layer 3-type VPN [virtual private network] products and also create a peering infrastructure, as we connect to and peer with other providers." (See Global Capacity Adds MPLS to Its Service Mix.)
In addition to the high-speed MPLS backbone running on top of the Ethernet core, Global Capacity will support applications that don't require MPLS such as MEF-based Ethernet applications, to avoid introducing the latency introduced by touching the higher layers of the network. That low-latency transport infrastructure will support special applications at speed up to increments of 10 Gbit/s, Thakur says.
On the access side of its network, Global Capacity is also looking to deliver higher speeds: "Ten meg is the new T-1," Thakur jokes. That will be accomplished both by putting in new infrastructure and leveraging what's already there, he says, including Ethernet over copper existing NNI [network-to-network interface] connectivity to its One Marketplace partners, who are tied together via an automated ordering process that allows faster identification of what connections are available where.
"Scaling up to support higher bandwidth on the access side of the network is another project we are working on," Thakur says. A lot of that demand is being driven by the movement of applications to the cloud and the movement of storage and compute to the data centers at the network's edge, he adds.
Global Capacity is getting involved in the virtualization drive as well -- something Thakur says he has already engaged with in previous jobs. The driver for NFV and SDN at his current company will be more efficient delivery of services and he says the early interest will be focused on practical solutions or "the technology that is closer to being ready" for edge services such as virtual firewalls or virtual CPE.
"We could deliver that on the edge with a single platform and the service itself whether it's a virtual CPE or firewall or enterprise session border controller," he says. "That does a number of things -- one of the things is speed to delivery of the service. It also helps with the logistics associated with managing the CPE program and it changes the cost structure of delivering those services. So we are starting to look at that now, in addition to what we already have on our plate."
Thakur sees his biggest challenge as increasing network speeds, scaling the infrastructure gigabit Ethernet and 10Gbit/s Ethernet, which will require a 100Gbit/s infrastructure. "So scaling is a big challenge and then, from an application perspective, getting the tools and systems in place, ready to manage applications over that network," he says.
Global Capacity is not yet engaged with any of the industry open source groups addressing the management and orchestration challenges, but could well join that effort in the future, he says. And while the carrier would love to see vendors deliver better tools for managing applications and services, it has its own internal core competencies in that area. At the end of the day, there is likely to be a combination of internally developed tools and those acquired from vendors, including potentially network orchestration and management systems. It will come down to what makes sense, in terms of cost and time to deployment, he says.
— Carol Wilson, Editor-at-Large, Light Reading