Growing adoption of cloud services has compelled hyperscale cloud providers to build global data center and backbone networks to support regional markets, with data centers positioned in important hubs all the way to the network edge. At the same time, regional cloud providers are building carrier-neutral and colocation data centers to meet more regional requirements including local performance, scale, and data sovereignty.

The demands driving these new data center deployments will ratchet up pressure on the interconnection infrastructure. For instance, the latest Equinix Global Interconnection Index¹ predicts that global interconnection bandwidth will grow at a 40 percent five-year compound annual growth rate (CAGR), reaching 27,762 Tb/s, which is equivalent to 110 zettabytes of data exchanged annually.

To keep pace with growing expectations, cloud providers need technologies that can boost the resilience, capacity, and operational performance of their interconnection infrastructure, while making it simpler and cheaper to operate. Fortunately, there are several approaches cloud operators can take within their optical data center interconnect (DCI). Design of optical mesh networks, staying current with the latest in optical innovation, and implementing automation can all play a big role in helping them succeed.

Improving resilience with an evolved optical mesh topology

Evolving existing 1+1 point-to-point DCI deployments to an optical mesh topology opens the door to cost-effective 1:N resilience and survivability capabilities that are essential for mission-critical networks. This evolution creates a basic mesh topology by connecting each data center location to at least two other locations.

Resilience is the main benefit of an optical mesh network. With a well-planned optical fiber mesh design, a network operator can diversify regional and international routes to establish more resilient connectivity. This means that if a primary data center link fails, the operator can rely on a secondary link to transmit data to the destination through a different optical route.

Cloud services generate massive amounts of data that need to be transported between data centers. They also require data backups using synchronous and asynchronous replication between multiple data centers, and often between regions to preserve business continuity and enable disaster recovery. If a link goes down, replication ensures that backup data transfers can continue over alternate optical routes with limited disruption.

Ramping up capacity and performance with new optical technology

Ever-increasing data traffic continues to drive the development of new optical technology aimed at scaling capacity without limits from the remote edges of the network all the way to the core.

For example, coherent optical technology offers modulation techniques that improve wavelength performance for wavelength division multiplexing (WDM). Through constant innovation and refinement, this technology has enabled massive scaling of speed in terms of bits per second (b/s), spectral efficiency (b/s/Hz) and capacity-reach (b/s-km).

This has taken wavelength speeds from 40Gb/s to 800Gb/s and increased all-optical transmission reach from hundreds to tens of thousands of kilometers. Continuing advances in this area will enable further scaling and pave the way to multi-terabit coherent optics.

WDM enables the parallel transmission of multiple data streams over many individual wavelengths across the optical fiber transmission spectrum. It has enabled network operators to massively scale their network bandwidth while paying off the high cost of deploying optical fiber by generating revenue over many channels.

While the WDM optical line systems of a decade ago typically supported 40–80 wavelengths in C-band spectrum, many of today’s optical networks operate in the C and L bands. This more their doubles their total fiber spectrum and data-carrying capacity. Future innovations in fiber and cable technology will enable operators to take network capacity even higher.

Reducing complexity with optical network automation

Optical technology innovations will increase speeds and reach, and architecture design aspects will increase resilience and survivability, but cloud providers must still find a way to scale their network operations. To keep pace in a highly competitive environment, they need to react faster to new demands and get maximum value from their optical networks. But as networks become more complex, it’s becoming much harder to reduce operating costs and increase agility. Automation can help.

Optical network automation uses centralized, policy-driven software to control and improve network behavior. This makes the network more responsive, efficient and reliable – and simpler to operate. With the latest innovations in network programming and mature open APIs, optical networks can be programmed to self-configure, optimize their performance in real time, and recover from failure events automatically.

Network automation can also enable cloud providers to:

• Deploy services faster, which accelerates time to revenue and improves the customer experience;
• Increase operational efficiency by reducing the effort involved in performing repetitive tasks and simplifying complex operations;
• Improve agility by rapidly deploying network updates and responding more quickly to changing demands;
• Boost quality by reducing human errors and system outages and delivering consistent, predictable outcomes.

Ready for the future with optical DCI

The demands placed on optical networks can be daunting. But cloud providers can face them with confidence, knowing that they can count on continuous innovation in optical technology to prepare their DCI infrastructures for the next generation of cloud services.

^{1. The Global Interconnection Index (GXI), Volume 6, Equinix, 2022}