By Ellis Booker
4m read time
Data networks are the core component behind enterprise digital transformation, connecting everything from distributed data centers to on-premise or colocation facilities, IaaS public clouds and SaaS services, and edge devices in the field.
Paul Savill, former SVP of product management and services at Lumen Technologies and now global practice leader of Network and Edge at Kyndryl, explains the limitations of traditional networking, and why there has been an industry-wide shift to software-defined networking (SDN).
Q: How have traditional networks made it difficult for enterprises to react to changing business dynamics and take advantage of multicloud/edge?
A: Enterprises that have been in business for quite a while have legacy network infrastructures, with fixed circuits that require manual design, provisioning, and turn-up. Provisioning and capacity planning on these networks takes a long time because you have to plan well ahead of delivery.
The other thing about traditional networks are the service interruptions that happen during reconfigurations. Most often, if you want to reconfigure a circuit that’s in a production environment, you have to create a maintenance window to plan and accommodate the potential for an outage. These maintenance windows can create a significant management burden on the enterprise.
Another big problem is that the hardware and software is very dedicated and very vendor specific. This creates operational problems because enterprises have to have people who are experts for particular vendors. Managing all of the nuances and technical differences between these vendors is expensive and difficult to support from a skills standpoint.
SDN helps create an abstraction layer above all this. Virtualization and SDN are really revolutionizing the networking world. It’s parallel to what happened in the world of computing 10 years ago, when compute and storage were tied to hardware from specific vendors. Virtualization came along, and along with cloud computing, enabled much greater flexibility and agility.
Q: Users worry about the complexity of managing their legacy network infrastructures while simultaneously migrating to SDN. Where should they begin this journey?
A: Enterprises are right to worry about it. Even knowing where to start is part of the problem because it depends on the nature of your business. A retail branch network is very different from a manufacturing network, where the tolerances for latency and backup are stricter.
This is where Kyndryl brings value, because we're helping build these solutions across the world, and learning from that. We don’t try to push a one-size-fits-all approach. We come in and help customers assess the types of networks they have, and what they need, based on their business model. How you approach the transition into the new world still involves a lot of customization. It isn’t a cookie cutter. It depends on the organization’s legacy architecture, what technologies they’re using, the SLAs and performance that’s required, and the kind of applications that will be running. Kyndryl specializes in taking all of that into account to craft a solution that is optimized for the customer’s real-world operating environment.
Q: What’s the significance of edge computing to Industry 4.0 and digital transformation?
A: Edge computing is going to become pervasive across all sorts of industries. I don't think it’s possible for us to make the leap into where we see things going without it. I often describe edge computing as opening up a third landing place for compute, storage, and applications. In the old world, you could either run applications or store data on-premise, or you could store data and run things out of a central location, like an enterprise data center or the public cloud. By opening up this third alternative, it creates new configuration opportunities for distributed applications that have advantages in overall cost, ease of operational support, and performance. Enterprises that can exploit these benefits can create competitive advantage in new revenue streams and business agility.
Q: What exciting opportunities do you see for low-latency edge computing?
A: There are many exciting low latency use cases we see emerging around the world in the industrial, financial, and retail verticals. Data-intensive industrial processes are often discussed, but one big box retail customer I dealt with was experimenting with a shelf-stocking robot, which used video analytics to evaluate the shelf inventory and identify when restocking needed to happen. They tried running the video analytics and the robotics control out of the public cloud, but the latency performance from the public cloud to the retail stores was so poor that they could not control the robot from there. So edge compute was where they landed. It gave them that balance of cost and performance.
For more information about Kyndryl’s Network and Edge practice click here.