The journey starts now for hyperscale capacity gains
18 September 2020 | Ben Baker
The timeline to hitting the 400GbE sweet spot is within range – and it’s time to prepare writes Ben Baker, senior director of service provider product marketing at Juniper Networks
In principle, it sounds like a perfect solution – the ability to quadruple the capacity of the fattest pipes most commonly used by data centres and WAN transport links. It offers the promise of unlocking new possibilities and enabling companies to do more future focused innovation right now. But the truth is that 400 gigabit Ethernet hasn’t seen widespread adoption quite yet. The good news is that the industry is making the right preparations to help a variety of businesses to benefit from this boost in scale.
Traditionally, when we talk about the companies who need 400GbE most, we talk about the cloud hyperscalers such as Google and Facebook. Their demand for data is growing dramatically all the time. But now, we’re seeing a perfect storm of 5G, 4K video and the recent boom in home working add more demand than ever. The use of 400GbE serves the need for increased digital services and content to be delivered more efficiently.
So, what’s stopping cloud providers, communication service providers (CSPs) and enterprises globally from deploying 400GbE immediately? It’s not the case that the technology is becoming a victim of its own hype. It can definitely deliver a host of revenue opportunities. But there are a couple of issues that the industry needed to address to give businesses the confidence to invest. Firstly, there’s the need for standardised implementations, and secondly, 400GbE pluggable optics must be available at the required volume and cost to make the tech work from an investment perspective. On both fronts, this is a battle that the industry is winning and as a result, there’s a growing sense of optimism.
Let’s talk about standards. There’s no getting around the fact that it takes a long time for vendors to agree on Multi-Source Agreements on optical standards. And even then, there’s the exhaustive testing that has to take place to ensure it’s fit for purpose. This process can be lengthy, but it does bring about benefits for the entire industry in the long run.
An example of this is the recent QSFP56-DD pluggable form factor which came with various benefits. It increases the bandwidth and density of optical transceivers when compared to optics that were previously available. It also offers DWDM and Ethernet in the same footprint, as well as offering backward compatibility with other QSFP formats. I suspect that in the coming months we’ll see more of this type of innovation and this will address the current shortfall of 400GbE deployments.
Making it pay
Then there’s cost. One telling stat will reveal a lot about why there’s a reticence in many quarters to invest in 400GbE on today’s economics. A decade ago, 10GbE optics made up around 10% of the cost of data centre networking hardware. Today, optics make up more than half. Although CSPs can make 400GbE solutions pay for themselves in the long term, there’s still a need for it to be cheaper to deploy and manage in the first place.
Here too, there are reasons to be confident about the future of 400GbE. The timely combination of bulk manufacturing processes, automation and economies of scale are allowing semiconductors to become more affordable all the time. These efficiencies will come to optical networking and make it more feasible to deploy 400GbE optics at scale. And as more hyperscalers deploy the technology, it will become more cost-efficient for everyone to do the same.
We’re fast approaching the tipping point at which 400GbE goes from a costly, niche pursuit to becoming a key point of competitiveness for a range of providers and enterprises. That means if you’re already in the networking planning phase, or are reaching the limits of your existing infrastructure, it’s certainly time to think about the role 400GbE can play in your future.
If the past few months have taught us anything about business planning, it is that you can never be fully prepared for what the economic environment can do to your day-to-day operations. But one thing we can be sure about is the growing reliance on cloud and digital applications and services and the need for them to be delivered with high uptime and resilience. With the timeline to hitting the 400GbE sweet spot within range, it’s time for businesses to prepare for what’s next.