Data Center Interconnect

OVERVIEW

The global Data Center Interconnect (DCI) market is exploding due to the emergence of large web-scale companies and rapid expansion of cloud-based services which is resulting in greater investments in new data center buildouts and higher capacity intra-data center infrastructure. As a result, there is a greater demand for moving terabits of bandwidth between data centers thereby driving the optical DCI networks. Besides traditional telecom carriers with pre-existing optical fiber networks, dedicated Internet Content Providers (ICPs) are also participating in this market. Emergence of high-bandwidth 4K video and 5G applications (especially Internet of things) are expected to further accelerate this trend.

NETWORK REQUIREMENTS

Bandwidth Scalability: Unlike traditional telco networks, DCI networks are highly dynamic in nature. There is a constant need to innovate and upgrade product capabilities in line with rapidly evolving customer requirements. For example, while 100G and 200G solutions have gone mainstream in today’s DCI market, 400G is already close to commercialization. The network has to be highly scalable and it should be easy to add capacity on the same platform and fiber infrastructure without the need for large investments or massive network rewiring.

Multi-criteria Optimization: A key characteristic of the DCI market is the heterogeneous nature of the networks being built in various geographical markets. While the universal optical transmission objective is to achieve the lowest cost-per-bit, other criteria such as power efficiency, equipment footprint, and day-one network capex are also critical to optimize total cost of ownership. It is therefore important for DCI products to be flexible enough to optimize across varying performance criteria for a cost-effective rollout.

Traffic Variability: Traditional static optical networks cannot react fast enough to the “on-demand” nature of service requests on the Customer’s data center network. There is significant variability and burstiness in traffic originating from different locations in the network. Moreover, the traffic pattern fluctuates rapidly both in time and space thus realizing optimal capacity planning a nightmare for the network operations team. DCI networks have stringent latency requirements that will be accentuated by  future 5G networks dominated by real-time video traffic.

Network Resilience: End users expect reliable, on-demand, high-fidelity access to applications, infrastructure, and other IT resources. This implies that the network has to be constantly alert to adverse changes in quality of experience and instantly “self-heal” through pro-active re-routing of traffic from congested or deteriorated links before an actual network failure. There is a need for constant “load balancing” in the network in order to ensure that the right data is delivered at the right place without over-engineering the network and without any stranded bandwidth.

Tejas Value Proposition

  • Scalable capacity from 10G to 100G/100G+ wavelengths combined with 80channel DWDM support in a compact form factor ensures that cloud data center infrastructure can be fully leveraged and monetized in a cost-effective manner. Disaggregated architectures for unmatched scalability in multi-terabit range.
  • OTN for granular provisioning of services at ODU0 (1 Gbps) with advanced ODUflex for onthe-fly resizing of bandwidth in ODU0 units to service on-demand service requests from larger clients
  • MPLSTP and OTN for low-latency, low-cost bypass at OSI Layer 1/2 compared to IP routed networks operating at higher layers; ROADM technologies for optical layer bypass (Layer 0) and for building cost-effective IDC connections. LTE for the end user to achieve low-latency, high data rate access to the IDC
  • Strong quality of service features such as i) faster centralized restoration of circuits/tunnels with multiple fiber-cut protection, ii) congestion-based service prioritization and early fault detection, and  iii) granular hardwarebased performance counters for real-time billing and monitoring of latency, jitter etc to prevent client churn and assure the best quality of experience
  • Support for GMPLS to enable distributed control plane operations between network nodes to lower latency of operations. Advanced SDNready NMS software to introduce greater network programmability with rapid setup and tear down of IDC connections in seconds