As the Asia-Pacific telecoms sector hurtles towards the coming digital economy driven by the Internet of Things and eventually 5G, the foundation for it all will be packet-optical technology, the cloud and virtualization.
That was the central theme of Telecom Insight 2016, a half-day conference hosted by Infinera in Kuala Lumpur last week that covered a wide range of technology topics – mobile fronthaul, business Ethernet, hybrid cloud and SDN – unified by the message that a next-gen optical infrastructure is required to support all of them.
Vinay Rathore, Senior Director of Solutions Marketing at Infinera, outlined the evolution of optical networks from now to 2020 and the market trends driving that evolution.
“In the past, 10G optical was driven by Layer 2/3 platforms, with network growth coming from enterprise and wholesale carriers,” he said. “Today, 100G is standard currency for the network and we’re seeing massive network growth driven by Cloud, Layer 2 & Optical. In the next stage, we’ll see a tectonic industry shift led by data centers, the cloud, the IoT and 5G.”
Rathore said the network will effectively collapse into two layers – optical transport and packets (or “Layer T” and “Layer C” respectively, as Infinera calls them) – with SDN to manage resources for both. The overall goal of next-gen packet-optical is to shed network complexity and port it to the cloud. Routers will be centralized and minimized in favor of a packet-optical infrastructure with WDM (which Rathore says provided the necessary scalability that optics will require) and SDN for resource optimization.
Jon Baldry, Infinera’s director of metro marketing, elaborated separately on how SDN can be applied in metro packet-optical networks: “Packet-optical distributed routing functions (DRFs) make the packet-optical network act like a router.”
Packet-optical DRF use cases include broadband and mobile backhaul, which minimizes backhaul cost and improves performance, and core router off-load/bypass.
“In the case of mobile backhaul, for example, without DRF all traffic from all base stations has to go to the core router, including traffic between edge routers and even traffic that could go between ports in the switch,” he explained. “Also, when you add new routers in the metro, that might require manual configuration changes. DRF enables more effective routing between routers because you don’t have to send it all to the core router, and DRF required limited configuration – any change in router configuration reflected in the DRF.”
Next-generation packet-optical also comes with a significant monetization opportunity in the form of Business Ethernet services (i.e. services defined under the MEF’s Carrier Ethernet 2.0 standard, including Enhanced E-Line, E-LAN, E-Tree and E-Access, as well as multiple classes of service).
Baldry said that packet-optical is ideal for Business Ethernet in terms of supporting on-demand bandwidth, efficient management and easy-to-use service provisioning. “But the dilemma for operators is how do you differentiate in a standardized world? One way is to combine services – for example, E-Access with E-LAN or E-Tree.”
The key thing to think about with Ethernet services, said Baldry, is this: “What if services suddenly ramp up? Or if it’s a mobile backhaul scenario, what happens when you shift from macro cells to lots of micro cells? You have to be able to scale up, and you have to be able to simplify it.”
Consequently, Business Ethernet requires an optimized packet-optical architecture that can scale fast and efficiently, as well as perform real-time SLA monitoring.
Cloud, 5G and IoT are key drivers
Vinay Rathore reiterated that the cloud will be a key driver of network evolution, although much of that activity is coming from content players like Google, Microsoft Azure, and Amazon Web Services, not the telecoms sector per se. IN fact, he noted, despite forecasts saying that 70% of businesses will be cloud enabled by 2020, telecoms giants like Verizon are getting out of the data center business, while companies like Microsoft and Equinix are investing billions in capital into the cloud and data centers.
“The point isn’t that companies like these are necessarily going to become your competitor, but you’ll probably want to partner with them,” he said.
Rathore said that for enterprises turning to the cloud, hybrid cloud is the most common and fastest growing choice – and it’s a target, not an intermediate step from public to private cloud.
“The thing is that you need tons of connectivity to do this, and the internet is not good enough,” Rathore said.
Previous options for cloud connectivity have included IP-VPN – which is an easy way to get started but security is questionable and performance is limited – and MPLS VPN, which offers better security and performance than IP-VPN but it doesn’t scale very well.
“What the industry is headed towards is an internet exchange [IX] model with dedicated access,” Rathore said. “That gives you the security, performance and scalability you need. The thing is, you need fiber access to get there. Copper is not going to cut it. So it requires investment in optical data center interconnect.”
This is a recent development, he added. “The data center interconnect market didn’t exist three or four years ago.”
Rathore cautioned that this isn’t an either/or decision between IP-VPN, MPLS VPN or direct-access IX. Microsoft uses an “all of the above” strategy for its internet, Office 365 and Azure services, for example.
With 5G and the IoT coming down the track, mobile fronthaul was also on the agenda, with the main focus on 5G’s latency requirements of 1ms.
Jacques Lebosse, technical director for Infinera’s APAC metro business unit, made the case that a mobile fronthaul network not only delivers sufficient low latency for 5G applications, but is also cheaper to build and operate.
Lebosse also outlined several deployment options for mobile fronthaul: (1) passive fronthaul (the most common option, but it requires colored optics in RRH/BBU and its management capabilities are poor), (2) active fronthaul (with un-colored RRH/BBU interfaces; it supports protection requirements and is key in situations with limited fiber availability) and (3) semi-passive fronthaul (essentially passive fronthaul with added management/monitoring capabilities that enables operators to wholesale their passive infrastructure).
Lebosse says it won’t be an either/or choice: “A mix of options is always required.”
But he did add that the industry is “moving towards an architecture where fronthaul and backhaul will be handled on the same platform and chassis, using the same management solution, which means cost-efficient use of network resources for multiple service types.”