We are witnessing an explosion of data worldwide. This is due in large part to a confluence of trends, including the growth of cloud computing and AI applications, demand for Over-The-Top (OTT) services, and the proliferation of data from various Internet of Things (IoT) devices made possible by 5G adoption. Emerging use cases on the horizon, such as autonomous vehicles and smart cities, are expected to boost bandwidth demand even further in the near future.
This data boom is driving the need for ever greater throughput, reduced latency, security and intelligence. In turn, this trend is fueling the growth of the data center interconnect (DCI) networks that link traffic between massive hyperscale data centers. In fact, the DCI market is expected to grow at a compound annual growth rate (CAGR) of more than 12% between now and 2030.
Yet, to keep up with these escalating data demands, hyperscalers and data center operators require greater technology innovation in order to improve the performance, reliability, automation, sustainability and cost-efficiency of DCI networks.
On course to carbon-neutral
Data center space and budget constraints mean that network systems are being concentrated into an ever-smaller footprint, increasing thermal load on networking equipment. This design progression translates into a continual necessity to decrease energy consumption and reduce heat transfer, both inside the data center and across DCI networks. The pressure to improve sustainability is further compounded by additional factors such as rising energy costs, shareholder and customer expectations, global initiatives like the UN’s Race to Zero Campaign, and power-hungry technologies like 5G.
At the same time, the demand for more data speed and bandwidth requires reliable, high-performance networks with greater capacity, lower latency and flexible scalability. In order to meet all of these demands and maintain competitiveness, data center operators must cut costs and streamline operational efficiency, even as they scale and upgrade their DCI networks.
With all the converging factors impacting today’s data center operations, we are seeing enormous change in the architecture, design and operation of DCI networks. At the center of this evolution is the continued disaggregation of optical network architecture. A modular, disaggregated architecture enables DCI network operators to implement transformational improvements, including multi-vendor configuration, pay-as-you-grow capacity, scalability, and open, programmable software to manage virtualized resource optimization.
DCI network transformation
The data center environment is necessarily dynamic by nature, demanding greater speed, agility and efficiency. Along with increasing disaggregation, additional technology transformations are needed in today’s DCI networks in order to further improve performance, reliability, flexibility and power efficiency.
– Extreme scale and performance
Demand for ever-faster data speeds has ushered in the era of terabit networking. To meet this growing demand, more resilient network infrastructure is needed to respond to rapid increases in data traffic. This requires advancements in optical transport technology to enable extreme performance and scalability for next-generation DCI networks.
As today’s data centers progress toward the terabit era, DCI network transponder technology is evolving to support higher baud rates and improved modulation techniques, enabling higher bit rates. For example, the new 1FINITY Ultra Optical System features a digital signal processor (DSP) capable of up to 135 Gbaud operations. This allows the transport platform to offer data rates of 1.2 terabits per second (Tbps), with future upgrade capabilities to 1.6 Tbps. This advancement in baud rate means that this next-generation transport platform achieves more than twice the speed of conventional products.
Likewise, new amplification advancements are also required to extend overall signal reach and capacity. With the introduction of Forward Raman amplification, optical signals can propagate in the same direction, reducing the influence of noise and extending the transmission distance. For example, the Forward Raman amplifier available with the 1FINITY L900 Series Optical Line System enables a 40 percent increase in either reach or capacity for a given fiber span.
– Enhanced sustainability
In previous ROADM networks, conventional transmission solutions were divided into C-band and L-band, resulting in wavelength range limitations. This was particularly problematic in metro and long-haul DCI networks. Evolution to continuous C+L ROADM architecture enables transport platforms to handle both wavelength bands at once, contributing to an increase in the maximum transmission capacity of the system. Moreover, with a solution able to support continuous C+L ROADM service from Day 1, such as 1FINITY Ultra Optical System, the transport platform uses only half the node equipment versus current bolt-on approaches, thereby achieving an 80 percent power reduction versus C-band only.
Yet, network operators are under continual pressure to improve sustainability, and power consumption is only part of the story. To address this issue, Fujitsu introduced closed-loop liquid cooling into the 1FINITY T900 Series Transponder — the first optical network transponder to use this advanced sustainable technology, which not only reduces power consumption but also provides better heat transfer to maintain a lower operating temperature for improved cooling efficiency. Moreover, a closed-loop liquid-cooled system offers longer fan life since fans run much slower, delivering higher reliability and greater sustainability, which translates into reduced maintenance.
Originally developed to handle the intense thermal load generated by the world-leading Fujitsu Fugaku supercomputer, unique closed-loop liquid cooling technology enables twice the cooling capacity and half the acoustic noise of conventional air-cooled systems. In the long run, advancements in sustainability will allow DCI network operators to increase network infrastructure scalability while contributing to the decarbonization of the entire optical network.
– Automated, optimized operations
Maintaining a live optical network historically has been a time-intensive, manual job. Conventional optical wavelength multiplexing devices required network operators to bundle multiple optical fiber cables within the device, resulting in high cable complexity and difficulties during installation. New smart fiber cabling systems drastically reduce troubleshooting time during installation and commissioning.
Furthermore, as DCI networks become increasingly disaggregated, operations in a multi-vendor environment are even more complex, increasing the need for automation and simplified architecture. With recent innovations in artificial intelligence (AI) and machine learning (ML), as well as automated performance diagnostics and fault localization, network operators can now fully automate system turn-up and optimization, decreasing human error, increasing reliability and cutting costs. As a result, overall network installation and commissioning is reduced from days to hours, enabling faster time to service with a lower total cost of ownership.
Hyper-reliable, hyperscale transport
As demand continues to heat up, hyperscalers and data center operators must scale and upgrade their DCI networks while keeping costs under control. This requires the latest optical technologies, next-generation architecture, and reliability that goes beyond metrics like mean time between failure to include extreme performance, automation, sustainable technology and a secure, global supply chain. This ‘hyper-reliability’ allows DCI network operators to stay competitive with increased performance, optimized capacity, greater sustainability and reduced cost/ per bit/ per kilometer.