The purpose of CORD
Network providers must address numerous hurdles with the demand for greater bandwidth and additional services. In an effort to tackle these challenges, operators hope to benefit from the economies of commodity infrastructure and the agility of scalable services. For these reasons, On.Lab proposed the idea of Central Office Re-architected as Datacenter (CORD) at the Open Networking Summit in 2015.
CORD is an architecture of the Telco Central Office (CO), which has a range of purpose-built devices assembled over 50 years. The aim of CORD is to make CO an essential feature of Telco’s cloud strategy, allowing them to provide more services on demand.
The architecture of CORD
The hardware behind CORD involves a set of commodity servers woven together by a fabric made of white-box switches. The switching fabric is used to direct east-to-west traffic flow. The racks of GPON OLT MAC substitute the closed hardware with an open software-defined solution. There is no north-to-south traffic flow.
Changing CO into CORD consists of two parts. The first part involves virtualizing hardware devices, thereby producing their software counterpart on commodity hardware. The second part involves providing a framework that the software counterparts can plug into in order to forge a unified end-to-end system. Software counterparts must fit together in a coherent way just as hardware-based CO fits together in a coherent way.
CORD integrates software-defined networking (SDN), network functions virtualization (NFV) and elastic cloud services. SDN allows network behavior to be centrally controlled using application program interfaces (APIs), or open programmatic interfaces like OpenFlow. NFV involves the decoupling of software from hardware; and the cloud allows network operators to innovate quickly with a variety of software-based platforms.
Benefits of CORD
By leveraging SDN, NFV and cloud technologies, CORD can lower capital expenses (CAPE) and operating expenses (OPEX). By virtualizing conventional hardware devices, network administrators do not have to invest in the installation and maintenance of physical equipment to run a chain of network connected devices. Instead, they can purchase inexpensive nodes to run virtual machines (VMs) that can perform a myriad of functions. These network functions can be installed on VMs in weeks. Installation on proprietary hardware, on the other hand, often takes months.
CORD can also provide quicker time-to-service for service providers. If an application running on a VM requires more bandwidth, for example, an administrator can dump part of the workload onto the shoulders of a different VM running on the same server. Resources are spread throughout the network rather than being confined to a limited data center.
Moreover, with the CORD architecture, third-party providers have a variety of ways to deploy VNFs. These providers can package their services in the shape of a VM or SDN control program. During the ONS conference, for example, ON.Lab was able to spin up a virtual CPE with a VM running on Openstack, and a virtual broadband network gateway application running as an SDN control program.
Conclusion
CORD is a major stepping stone for lowering costs and increasing agility for the telco CO. By combining NFV, SDN and the cloud, CORD can reduce CAPE and OPEX, provide quicker time-to-market for service providers, and offer third-parties multiple ways to deploy VNFs. Although still in the developmental stages, the architecture will continue to mature and grow as its marches toward commercial deployments.
