Let’s start by getting this out of the way early. Yes. The Nokia 3310 was a style icon and of course, the name ‘Nokia’ will bring out retro–gadget melancholy in all of us – from the snap-on covers to the ground-breaking polyphonic ringtone chip (designed by synth-pioneer Thomas Dolby), to the endless hours of playing Snake in glorious monochrome LCD.

Now we’ve got that out of the way, it’s 2023, so let’s move on, just as Nokia has. Hardik Gohil, Senior Director of Strategic Business Development for IP Networks at the company is here to explain why, although the phone in your hand might not have Nokia branding, every call and internet search you do is touched by the company. But as Hardik explains, shifting perception from a fondly remembered consumer product line isn’t easy.

“Every time I go to a meeting, I end up spending five minutes discussing how the Nokia 3310 was such a great device and how we used to build good phones – which is great. But the networks that we are building now? People are consuming them every day. This video call is being delivered on a network that is touching Nokia equipment.

“While we don't always realize it, Nokia delivers the network that enables us in modern life. We've been supporting some of the leading hyperscalers and large enterprises in the world. If you've landed at an airport, or you've used a train to go from place A to place B, some part of it has been because our routers, switches, and equipment were working perfectly as intended.”

“It's not just about Nokia delivering the network, we're also supporting some of the world's most mission-critical enterprises for their network needs, and we've been doing this for a long time. We support public utilities, and public safety networks to keep their communications running smoothly. We deliver networks that enable you to browse the internet 35,000 feet in the air. As well as data center infrastructure, we ensure that data centers interconnect with each other. We are one of the very few companies that can give you a complete solution.”

Speeding up data transfer

He’s right, of course. These days, Nokia's infrastructure is everywhere, reflecting a company that has always evolved with the times, from rubber manufacturers at the turn of the 20th century, to cables, to set-top-boxes, to consumer handsets. Nokia has always pivoted towards the future. So, with AI as the latest revolution to come knocking on their door, the company is focused on modernizing and automating the enterprise data center to deliver the scale, capacity, and performance needed to manage these applications and their huge workloads.

“There's no doubt that AI and machine learning is going to change the way the industry is, as we know it today.” explains Hardik, “But a common characteristic of these AI workloads is that they are very data and compute-intensive, so they leverage massive data lakes to make better models. And for doing that they take those billions of parameters and distribute them across DPUs, CPUs, and GPUs. These high-performant processors compute, exchange data, recycle, and then the next cycle begins. Approximately 20 to 25 percent of the time is just spent exchanging data, so networks must be efficient, lossless, and have higher capacity because any bottleneck in a sub-optimal network would lead to a substantial impact on job completion time.”

Efficient transfer certainly makes sense as a pinch point. But the secret is not just about efficiently transferring everything, it’s also about cutting down the latency to as close to zero as possible.

“Having IP networks with scalability and flexibility always helps, and higher capacity always helps. But it's not always true that higher throughput means anything for AI, it's about throughput versus goodput. You need all the throughput, you need all the data to be reliably exchanged, but you also need it to be delivered in a lossless manner. When it comes to improving networking capabilities, you can enhance the data center, and build it in a way that caters to terabit routing capacity, and high-performance, lossless communications. You should be designing a data center, or even if you want to upgrade a data center – to be flexible, scalable, lossless, and low latency.”

Automating to negate

Machine-to-machine is all well and good, but humans will always be a part of the data center and can often be a pinch-point. However, the advent of AI-enabled systems has further advanced the number of day-to-day tasks that can be automated. Hardik tells us:

“When you hear a bad news story about an enterprise, it usually comes down to manual errors or something that could have been avoided. That’s why automation is key to ensuring tasks can be done, but it's a journey. It's not that everything is ready right now, but still, we need to embark on that journey. It also allows you to make sure that human resources are dealing with the needs of the business, rather than spending their time doing repetitive tasks.”

AI is just the latest evolution in process automation, though a timely one in the current skills shortage crisis:

“When we talk about automation, it's a more generic term than talking about AI. Automation means automating manual routine mundane tasks, to streamline them, making them better, and making data center operations more efficient. We are scaling data centers to cater to the needs of the business. But we cannot scale resources at the same level. We don't have the same expertise everywhere, so we need to empower our enterprises to work with scarce resources. How do you do that? You automate to keep your efficiency to the maximum while reducing errors.”

Open and transparent

At the heart of Nokia’s approach to streamlining data centers is its dedicated SR Linux Network Operating System, or NOS. As the name suggests, SR Linux is an open-source operating system, built from the ground up to provide all the things that data center operators crave most. Hardik tells us what is at the heart of SR Linux:

“Openness is the term that people use a lot. We need these kinds of open NOS to unleash the creativity of your DevOps, and your NetOps community. You need openness so that you offer flexibility for enterprises that need to cater to their business demands. You want the data center to focus on delivering the needs of the business and not just be a switching layer. You want your network to give you as much visibility as possible so that you can decide what the best way of going forward is, using all the available data. You need a trusted and stable platform where you and your applications are secure. The next generation NOS should be open, agile, and simpler to operate as well.”

That theme of visibility – the ability to find everything you need in one place, is something that Nokia recognizes as key to the success of any data center facility.

“You want to know what's happening in your data center, and that it is performing at the level that is the most optimal for your applications. So, visibility is important, not just at a high level, you need visibility at all kinds of levels to make sure that you can make the right decision. We can diagnose the problems faster, we can speed up decisions, and eventually, it results in higher productivity of your application.

“Imagine a scenario when you have an AI workload and you're seeing things aren't happening efficiently. Visibility at that layer to understand why that is happening is critical because if you don't have that, you're using expensive GPUs and DPUs, and if they are underperforming, or less efficient, your investment is not being recovered.”

The repatriation wave

This is a vital factor in enabling hybrid workloads because, after an emerging preference over the last decade of moving data to the cloud, we’re now seeing a steady increase in migration back to on-prem and Edge models:

“What enterprises have now found is that we need to do some rebalancing. Some applications are ideally suited for the cloud. However, some applications are best suited in their own physical environment in their own space. That “repatriation” is the theme for the next few years, we must find the right applications in the right places, and then we need to optimize and automate them. We also must figure out how to get those applications that never moved, out of the legacy infrastructure and get them into the cloud. We still have some ways to go. Enterprises are still trying to figure out what's the right place for the right application.”

The idea of repatriation rings alarm bells for Hardik because operators are not always considering the capacity required to bring these applications back from the cloud:

“When these applications are moving back from the public cloud for whatever reasons, the data centers are not in shape to handle the cloud-native applications that are coming in. So when we build our data centers, we also have to account for the fact that we will see that repatriation coming in. You always must look at least having one eye towards the future, and then also think about the current.”

Glass and sand

Nokia has taken this into consideration in designing its NOS, which comes with the holy grail, a single ‘glass cockpit’ providing complete visibility across the entire network:

“At the end of the day, management is looking at your complete network from a single pane of glass. You're going to land in a situation where you will have a data center on-prem, you will have it in a public cloud, you will have it on the Edge. As a data center operator in an enterprise, I want to look at everything on a single pane of glass. That is key. So what we need to do as vendors is to make sure we give them enough to support a single pane of glass management to make sure you have full visibility of your network.”

In spite of this, there’s no suggestion that cloud-based applications are anything other than here to stay, and Nokia has taken this into account in SR Linux, which takes the concept of the ‘sandbox’ to a whole new level:

“When we were designing our NOS, we needed to have toolsets to make data center operators confident in upgrading the data center at speed. So we came up with the idea of a digital twin, a data center in the metaverse. That means you can test your changes and roll out your upgrades on that virtualized copy. It takes data from the live servers, so it’s as close to the real data center as possible. What we have to do as vendors is to allow our operators to easily and quickly upgrade with confidence. We want to make sure that whenever they touch their network, they do so with confidence and they know that this is a reliable mechanism.”

A partner for the journey

Ultimately, SR Linux is just one element of the partnership that is forged between Nokia and its clients. Hardik speaks with passion, not just about Nokia’s products, but about the entire lifecycle relationship that doesn’t end with installation. Rather, that’s where it begins.

“There is no play for us if we don't give you support. For us, it's not just about selling the product, it is going through that entire journey with you, deploying it, and then supporting it. Because it's a journey, right? We've worked with our customers to figure out the problems that need to be solved. We focus on three kinds of applications: business-critical, mission-critical, and society-critical. We're looking at solving problems for people. We’re not looking at simply building the best products, at the end of the day, we’re looking at solving problems. You need to make technology more consumable, and more friendly for users and businesses. We are continually striving to make the adoption of cloud and data center networking technology as easy as we can.”

For more information on Nokia's enterprise cloud network capabilities, click here or check out the video here.