In March 1972, at the dawn of the Internet, the Club of Rome published “The Limits to Growth.” This landmark report was the first to model our planet’s interconnected systems and the limitations of its natural resources to support the needs of a fast-growing global population.
Today, 50 years later, the Internet has become the global fabric of a fast-growing digital world. It plays an essential role in interconnecting humans and machines on a global scale and kept us going during the Covid pandemic by moving bits instead of bodies. Telecommuting is taking many cars off the road, but the energy required to fuel the exponential growth of data centers and IP network equipment is significant. Estimates are that the cloud’s share of our carbon footprint will grow from 3.8% in 2020 to 8% in 2030, and sustainability has become a key concern for network operators.
Data traffic increases annually at a rate between 19 and 28%, which accumulates in growth between 5.5 and 13 times over 10 years. Hyperscale data centers deep in the core cloud can be built near hydro plants, in remote areas, and in cooler climates to save power, space and cooling. The challenge is at the IP edge and in the IP core network as there are practical limits to the available power, space and cooling in central offices and equipment racks. Sustainable net zero traffic growth, therefore, requires that any increases in IP network capacity are offset by an equivalent increase in resource efficiency.
Scaling IP network capacity in the same footprint
800 Gigabit Ethernet (GE) routing technology ranks high on the agenda of many IP network operators because it enables them to meet the growing demand for bandwidth while simultaneously decreasing the space and energy needed by routing equipment. Currently, most IP networks max out at bit rates of 100 to 400 Gb/s. Increasing capacity then becomes incrementally more costly and cumbersome to manage because IP traffic must be distributed over multiple links and aggregation groups. This gradually consumes more routers and interface ports that use up precious space, energy and cooling resources.
800GE routing is the next generation of IP networking technology that enables operators to upgrade router interfaces with 800Gb/s pluggable optics. It leverages the most recent interworking specifications of the QSFP-DD MSA group for an 800GE interface connector with eight electrical lanes using 112Gb/s signaling rates. The new 800G QSFP-DD connector has similar dimensions and is backward compatible with 400GE and 100GE QSFP-DD pluggable optics.
Using 800G QSFP-DD line cards, operators can gracefully scale port capacity from 100G to 200G, 400G and 800Gb/s in the same chassis and link topology, by simply plugging in faster optics. Alternatively, an 800G QSFP-DD port can connect two 400GE or up to eight 100GE ports using a single 800GE pluggable and an optical break-out panel, which greatly increases faceplate port density and simplifies cabling.
Besides offering more capacity for traffic growth in the same rack space or chassis, 800GE routing also reduces the energy consumption of router ports and pluggable optics. While 800GE pluggable optics cost about the same as two 400GE pluggable optics, they consume between 25 to 42% less energy per bit because relatively fewer lasers and less circuitry are involved.
Getting ready for 800GE routing
800GE routing drives new requirements for faster, more integrated and power-efficient routing silicon.
- The faster 112Gb/s SerDes signaling rates needed for 800GE pluggable optics impact the chip-to-chip interfaces of packet processors on interface modules (line cards) and switch fabric modules.
- 800GE routing silicon must have at least double the capacity and twice the density of 400GE silicon and be at least 50% more energy efficient to fit in the existing equipment resource footprint.
- The thermal performance of line cards and chassis must accommodate at least 25 watts per port for the evolution to 800GE digital coherent optics.
Modular routers with traditional backplane and midplane chassis designs are typically unable to reliably achieve 112G SerDes signaling rates and thermal performance for high-density 800GE interface optics. Modern switch fabric designs use an orthogonal direct cross-connect fabric or fly-over cables to support 112G SerDes signaling paths while allowing optimal front-to-back airflow through the chassis that is unimpeded by a backplane or midplane.
While many operators are still transitioning their IP networks from 100 to 400 Gb/s link speeds, traffic keeps growing relentlessly and 800Gb/s is increasingly important to scale capacity further.
800GE routing offers a smooth upgrade path from 100 or 400 to 800 Gb/s link speeds, with immediate and recurring operational savings on energy costs and rack space. While there is a one-off capital expense to upgrade IP routing platforms, there are recurring operational cost penalties in deferring the inevitable investment in 800GE capabilities.
Besides consuming more energy and space, continued investments in legacy 100GE and 400GE routing platforms that can not be upgraded to 800GE may result in a more costly transition with a bigger write-off on capital investments later. In conclusion, 800GE routing technology offers IP network operators a more sustainable path for future traffic demand with lower operational expenses and better hardware investment protection. It’s good for your business and good for our planet. Double win!
