Mission

Search

Aggregation at the Edge: Beyond the Decentralized Versus Centralized Debate

Play episode

Many times we think about ideas in terms of debate — that there are only two ideas that are oppositional in nature and that one is superior to the other. This is true right now in the network conversation concerning an industry push toward either decentralized edge computing or more centralized network systems. What if, as is often the case, there is another way to think about duality where a unity lies somewhere in between? Raj Yavatkar, CTO of Juniper Networks, believes that to be the situation in terms of the debate for supremacy between decentralized versus centralized networks. 

“So before I get into the debate, which is a very interesting debate, you can say that three factors, a trio of trains, are coming together. People are putting more and more workloads to the public cloud, the 5G infrastructure, which is very densified, and there [are] micro cells.There’s a densification of the traffic, which is at the edge. And third is the new applications based on A.I., M.L. and analytics. So when you look at all these three trends, it’s not so much just sending the traffic to the public cloud. You’re using edge from multiple lists.” 

In this episode of IT Visionaries, Raj proposes that, in truth, networks can be both things simultaneously as they become increasingly “aggregated at the edge.” He also shares that teleportation via quantum networking is occurring right now.

Main Takeaway

  • Aggregation at the Edge: Rather than an either or proposition concerning decentralized edge computing or centralized network consolidation, there is a great deal of network aggregation at the edge. On one hand, there is a drive to centralize due to the ability to send traffic to the public cloud. Alternatively, issues concerning limits of data transfer across national boundaries due to GDPR regulations as well as the potential need for local networks for automation are promoting edge computing. In reality, aggregation is still occurring but simply on the edge.
  • Customers Drive Tech Advances: Customers are the vanguard of technological advancement and driving development based upon their evolving needs. Customers increasingly want automation and predictive analytics to support maintenance. In terms of network service, this includes maintaining the physical products that support the network operating seamlessly.
  • Quantum Networking is Real: Not only is computing capacity advancing rapidly but so is communications. Simply put, quantum networking is real. It is based upon the idea of encoding data into quantum bits in order to send it along a network. In other words, teleportation is not relegated to the realm of science fiction. It’s happening now.

For a more in-depth look at this episode, check out the article below.


Article 

Many times we think about ideas in terms of debate — that there are only two ideas that are oppositional in nature and that one is superior to the other. This is true right now in the network conversation concerning an industry push toward either decentralized edge computing or more centralized network systems. What if, as is often the case, there is another way to think about duality where a unity lies somewhere in between? Raj Yavatkar, CTO at Juniper Networks, believes that to be the situation in terms of the debate for supremacy between decentralized versus centralized networks. Yavatkar proposed that, in truth, networks need to be both things simultaneously as they become increasingly “aggregated at the edge.”

“So before I get into the debate, which is a very interesting debate, you can say that three factors, a trio of trains, are coming together,” Yavatkar said. “People are putting more and more workloads to the public cloud, the 5G infrastructure, which is very densified, and there [are] micro cells.There’s a densification of the traffic, which is at the edge. And third is the new applications based on A.I., M.L. and analytics. So when you look at all these three trends, it’s not so much just sending the traffic to the public cloud. You’re using edge from multiple lists.” 

There is no debate that media consumption is ever increasing — today more people use popular applications to watch movies, play video games, connect with loved ones, and to work. Yavatkar knows this, and he explained that media consumption is growing right now, due to an uptick in remote work over the last 18 months, as well as an increase in bandwidth.

“So that’s also really fueling this huge demand for consumption and that’s across European countries,” Yavatkar said. “You’ll be surprised to know that one of my friends is the President of Zoom. We are using Zoom here. And when he travels to India, he says, even in a small village, people know what Zoom is.” 

Although there isn’t a debate that media consumption is expanding — and with it the need for network systems to support these increases — there are different opinions as to what sort of networks best serve this aim. Some industry experts suggest network decentralization with more localization as the most prudent approach, while others offer that centralized networks are the best path forward. Yavatkar dispels that networks will either be decentralized or centralized; that they should be local or global. Instead, Yavatkar suggests network traffic must become “aggregated at the edge.” He cited regulations and the needs of some companies for local network control as two reasons traffic is consolidating at the edge. 

“There are data sovereignty, GDPR-like regulations, which require in some countries that you cannot send the data to the public cloud all the way if it is crossing the national boundary,” Yavatkar said. “That also is making sure that edge is important. And applications are moving there because if you look at applications like smart manufacturing, I talked to companies like Ford, BMW, or Tesla, they have robots to do lots of the manufacturing and the control has to be local. It cannot be sent all the way through to [the] public cloud.” 

From there, Yavatkar stressed that network customers are propelling new technologies and innovations. He expressed how customers desire more automation as well as predictive analytics for maintenance. As an example of the usefulness of predictive analytics, he gave the airplane industry as an example.

“They don’t wait until the engine fails,” he said. 

This is also true for maintaining networks.

“Why can’t we apply [a] similar principle here?” Yavatkar said. “Through predictive analytics, you know what rate that cables and switchboards fail. So predictive analytics will allow you to do proactive maintenance, including automation of RMA. That means the replacement of parts you ordered [are] there. You don’t worry about supply chain constraints. You account for that, and the part is there, and you are ready to replace it even before it fails.”

Yavatkar also suggests that many customers now want a data center provided to them as a service. Furthermore, they want to be able to set up and operationalize these data centers quickly via an easy-to-use interface. 

“[Customers want] data center as a service,” Yavatkar said. “Don’t tell me how you bring up the data center. You do it yourself and offer me that as a service so I can simply make an API call or make a click of a button and user interface. And I get every data center set up very quickly and start operationalizing.”

While customers are definitely pushing for new technologies, it is also true that the rate of technological acceleration is just as profound in computing as it is in communications. Where computer power doubles every eighteen months, Yavatkar points out that communications power does that every twelve months. Since this is the case, the future of communications is very exciting. 

Yavatkar closed the conversation detailing the concept of quantum networking — where data is encoded into tiny bits and then sent along a network. From his point of view, the technology no longer lives within the constraints of a customer’s imagination. 

“By the way, it’s not science fiction,” Yavatkar said. “Today, that technology is coming out. Today, it’s in defense labs [and] on some startups. I can point us to start. But [teleportation] is going to happen. Quantum networking is going to happen before quantum computing happens in that sense, because quantum computers are being built. They’re getting bigger, and bigger, and bigger. But quantum networking through this teleportation is happening today.”

Once a notion relegated to science fiction, teleporting is here. If that is the case, then the future we once imagined is happening now and, the question remains, what will then be the future? There’s no doubt it will be driven by consumers, customers, and by innovators such as Yavatkar. 

To hear more about how Yavatkar and his team are connecting the world, check out the full episode of IT Visionaries!

 


To hear the entire discussion, tune into IT Visionaries here

Menu

Episode 304