Ever-evolving broadband services and the promise of 5G wireless technologies are forcing service providers to deploy an increasing mix of wireline and wireless services in their networks. In many cases, this evolution will drive the need to bring service delivery closer to the edge of the network. As a result, operators are being forced to incorporate datacenter functionality into their central offices alongside traditional telecommunications services.
This convergence of datacenter and telecommunications services will push different service provider groups to work together. But supporting low-latency data applications and ‘standard’ telco services in the same facility requires developing two different mindsets: operators need to support the ’rip and replace‘ datacenter outlook while at the same time supporting evolving needs of the traditional central office infrastructure for a long lifetime.
Moving Data Centers to the Edge
As the network evolves, operators deploy multiple networks: fiber-to-the-home (FTTH), fiber-to-the-business (FTTB) and fiber for wireless networks, for example. There are different uses for these networks, one being for low-latency services. Latency is driven by the number of electronic conversions made – the number of hops the signal has to take to retrieve or transmit data. Each time a conversion is made, latency is added. But 5G network applications can’t afford many conversions, so 5G is driving the need for edge data centers closer to users.
"The convergence of datacenter and telecommunications services will push different service provider groups to work together"
For an operator, an edge datacenter will most likely be located in a central office. In a central office, the operator will have the traditional services it delivers today (FTTH, voice, and video) and will also host edge datacenter functionality. Edge data centers connect with regional data centers: edge datacenters will house applications that require the lowest latency, while some applications will go into regional datacenters. Regional datacenters would house less popular video programs, e-mail servers, SMS servers and other applications where minor delays in access are not critical.
Datacenter architecture and standards and methods of operation are very different from central office architecture and standards. Datacenters typically work on a three-to-five-year “rip and replace” cycle, while central offices work on a 10-20-year equipment lifecycle.
Coming into the central office, the operator will have to deal with a massive fiber network. The operator needs to have that fiber terminated in a high-density fiber distribution frame that offers easy access, has a lot of flexibility, is extremely reliable over a span of 20 years, and can support multiple network evolutions. As they serve the edge data center side, the equipment has to use multimode fiber and have a reasonable price (assuming a 3-5-year replacement cycle). Density and accessibility are critical, but so is long-term reliability.
While telco-oriented central offices have developed a lot of expertise in single mode fiber management and connectivity, adding datacenter functionality will require many more multimode connections. As a result, the telco needs to develop multimode fiber management and connectivity expertise.
To support data center-enabled services, the telco must also develop expertise in virtualization. The days of using numerous individual network elements for performing a single network functions is long past, and operators are maximizing their equipment spending by virtualizing network functions into software on servers and spreading applications across them in the form of workloads using technologies like network function virtualization (NFV) and software-defined networking (SDN). These techniques allow operators to put servers and switches into their networks; what might have taken 10 racks of dedicated physical network function equipment can now be handled with three or four cabinets. Virtualizing functions is more cost effective, uses far less space and lowers costs.
While it’s not possible to see too far into the future, there are three keys for future evolution:
• Plan for flexibility – Use of multifiber push-on (MPO) connectors for fiber cables and patch cords makes it much easier and more cost-effective to change configurations when needed. Panels that include modules that easily enable changes from LC to MPO and back to LC all while utilizing the same backbone cable is critical with the continued evolution of multi-mode optics used in data center servers and switches
• Plan for density – Operators should choose the highest-density fiber platforms and switching equipment to allow for future growth in connectivity for service delivery. They should also consider using wavelength division multiplexing equipment to scale capacity in existing fiber networks.
• Plan for accessibility – Operators should use fiber panels and frames that maximize access to fiber connections.
In the end, forecasting demand is challenging. The mix of current and new technologies factor into the equation. A flexible and adaptable infrastructure that allows providers to quickly tailor their services to customer demands will prove essential to creating a successful plan of action.