Data loads are getting larger, data ecosystems are becoming more distributed and data users are demanding real-time performance. For the enterprise, this is the triple-play of bad news because it puts the pressure on legacy network architectures to function in ways that their designers could scarcely imagine.
But this is also why many organizations are turning to fiber optics, not only for long-haul connectivity to cloud and colocation providers but across campuses and inside the data center all the way to the rack level. No matter what the use case, fiber is emerging as a more efficient option than traditional copper wiring, and the only one that can future-proof enterprise infrastructure against emerging data-heavy applications like artificial intelligence and advanced analytics.
In a recent post on Seeking Alpha, LD Investments laid out the case for fiber optics in the emerging digital economy. In the first place, cloud computing, mobile technologies and the IoT are on the upswing and will continue to accelerate bandwidth requirements of most enterprises well into the next decade. In the US alone, cloud traffic is expected to jump from about 2,300 exabytes today to more than 3,500 exabytes by 2020. That will require terabyte-scale carrying capacity that, frankly, cannot be accomplished in an economic fashion without fiber.
Meanwhile, mobile data consumption, led by a continuing upsurge of video traffic, will climb from 8% of total internet traffic in 2016 to 20% by 2021, while at the same time the IoT universe will top 30 billion devices by the end of the decade, with just one connected car generating about 25Gbit per hour. All of this data will increase pressure on 5G networks, which will have to rely on fiber backbones to handle the load.
In the enterprise, data infrastructure is evolving along dual paths, both of which lead to increased use of fiber. On one side is hyperconverged infrastructure (HCI), in which multiple nodes of compute-storage are networked together using mesh fabrics in support of abstract, software-defined data architectures. Already, organizations are planning for 10G, 40G, 100G and larger fabrics, which can certainly be built on copper but would provide even greater scalability with fiber. Even without HCI, the transition from traditional spanning tree and three-tier architectures to leaf-spine topologies is prompting many organizations to go all-fiber.
The other path has organizations scrapping most, or all, of their internal data infrastructure to pursue hybrid or all-cloud environments. This is leading to the gargantuan, scale-out data centers that will serve clients over regional, national or even international footprints. As Sustainable Business Forum’s Ryan Kh notes, this is leading to increased fiber dependency within the data center itself – and some of the newest facilities span acres – but also over the wide area under the data center interconnect (DCI) model. This allows hyperscale providers to not only build faster, wider connections to customers, but to other providers as well in support of applications like geographic load-balancing and high redundancy.
All of this is leading to a radically new data infrastructure in which the speed at which data is transmitted from place to place is replacing both processing performance and storage capacity as the enterprise’s top priority. As ADVA Optical Networking recently noted in an opinion piece for Knect 365, the new data center deployment model encompasses not only centralized data facilities and content delivery instances, but new layers of edge and on-ramp infrastructure as well. With the goal being to increase responsiveness to myriad end-points, some of which are tasked with life-or-death functions, organizations of all types and sizes will be hard-pressed to justify anything but the most advanced networking technology.
Fiber may not encroach upon every last segment of the data ecosystem, but for critical enterprise applications, and many non-critical ones as well, the advantages of high-speed optical connectivity are becoming too hard to ignore.