Kevin Wade, director of product marketing at Cyan (, says:

Large and mid-size data center operators are scaling their businesses by expanding to new sites and delivering more cloud-based services, two trends which are both leading to a dramatic rise in inter data center traffic. Expanding their WAN bandwidth to keep pace with this growth traditionally involves purchasing more leased-line capacity for data center sites, but this solution can be costly. Addressing this challenge becomes more of a priority as the number of data center sites and bandwidth requirements increase.

A new-generation of affordable and manageable “packet-optical” transport platforms, coupled with the availability of dark fiber, are leading a growing number of operators to deploy their own data center interconnect (DCI) networks. In addition to reducing or eliminating WAN leased-line costs, this approach gives data center operators and cloud services providers a more scalable network that they can control from end-to-end, allowing them to manage rapidly increasing bandwidth requirements more seamlessly.

Ethernet and Dense Wave Division Multiplexing (DWDM) are data center managers’ technologies-of-choice for DCI network architectures. Ethernet is preferred due to its familiarity and inter-operability with the data center LAN, while DWDM optimizes fiber utilization and provides latency and distance advantages.

The two primary challenges of deploying Ethernet over DWDMDCI networks are scalability and manageability. Specifically, standard L2 Ethernet and VLAN switching does not provide the scale and reliability required for data center interconnect applications. Ethernet switch/routers also do not incorporate optical transport functions, which would require the deployment of separate DWDM platforms. In addition to consuming more space and power, using two separate platforms makes management of the data center interconnect network, and capacity provisioning, a complicated and time-consuming process.

The challenges associated with DCI network scalability has been addressed in a new generation of packet-optical transport platforms (P-OTPs). P-OTPs scale Ethernet networks by integrating non-blocking1/10 GbE L2 switching in addition to standards-based connection-oriented Ethernet (COE) transport. COE utilizes IEEE 802.1ah Provider Backbone Bridging (PBB) and 802.1QayTraffic Engineering (PBB-TE)to scale virtual LANs by encapsulating MAC addresses within MAC addresses and by enabling the engineering of explicit Ethernet paths that deliver deterministic bandwidth and carrier-grade, sub 50 millisecond protection switching. COE is the ideal technology for implementing large-scale, resilient Layer 2 domains across multiple data centers.

P-OTPs also collapse support for multi-degree ROADM (reconfigurable optical add-drop multiplexer) and high-capacity DWDM transport. These capabilities provide dramatic increases in per fiber scalability and enable ultra low latency Ethernet transport.

Innovations in multi-layer management also overcome DCI manageability challenges by simplifying network implementation, monitoring and operations. Multi-layer network management incorporates object-based intelligence providing awareness of network resources such as switching/transport nodes, connections, and network layers, as well as their inter-dependencies, to optimize the efficiency of network planning and operations. Multi-layer network management also provides advanced 3-D visualization that delivers intuitive, graphical views of the network and network resources as well as “virtualized” views based on tenant, service type, region or other factors, allowing data centers to partition network visibility in a manner that is consistent with their operational procedures.

P-OTPs that implement connection-oriented Ethernet and DWDM/ROADM capabilities, combined with multi-layer network management systems, like those available from Cyan, allow data center operators to economically deploy and manage their own high-capacity data center interconnect networks over dark fiber. In addition to reducing or eliminating WAN leased-line costs, this architecture delivers superior Ethernet transport performance and unprecedented scale across multiple network layers, while simplifying network management and operations.