TPU Inference Servers for Efficient Edge Data Centers - DOWNLOAD

Edge computing shows centralized model of networking is broken; how to fix it?

Categories Edge Computing News  |  Guest Posts
Edge computing shows centralized model of networking is broken; how to fix it?

By Khalid Raza, CEO and founder of Graphiant

Edge computing is a class of distributed computing that brings data storage and processing closer to the devices and sensors that generate it. Edge computing brings with it advantages that include reduced latency and increased efficiency.

Not surprisingly, edge computing has enterprise IT’s full attention. In fact, analysts predict enterprises will spend nearly $1 trillion on edge computing in this decade on use cases such as Industry 4.0, Agriculture 4.0, telehealth, and autonomous vehicles.

Yet, edge computing also has two core requirements that are exceptionally challenging for IT: performance and security.

  • Much of what’s happening at the edge involves machines, not people. Factory floor equipment, autonomous vehicles and remote surgery gear generate tremendous amounts of data that must be processed quickly to make timely decisions. For example, in the near future, a surgeon performing a remote surgery may see an important artery in an unexpected place. She deftly moves her scalpel to avoid nicking the artery. It is imperative the edge network gets that command to the robotic scalpel in near real-time.
  • Security and Privacy. When data is generated — and consumed — at the edge, there is no time to backhaul data to the data center or cloud. This means the data must be secured locally.

In practice, the metrics that define edge networking performance are throughput, scale and latency. Throughput and scale need to be predictably high, while latency needs to be low.

Old networking models are broken

Networking was originally built on a centralized design. Data and compute power sat at the data center; that has been shifting over time. Cloud was the first change, but in most cases, the cloud simply acts as the new data center. Apps and data are centrally stored in the cloud.

With a centralized model, IT would backhaul traffic from remote users to the centralized location. That, however, won’t work with edge computing. The data is generated and consumed at the edge; backhauling destroys performance and adds latency.

Second, IT has struggled to keep everything secure as networks now include multi-cloud, work-from-home employees, external partners, customers and edge networks. It wasn’t too difficult with MPLS (Multiprotocol Label Switching), which provided solid security and performance as a private network. But MPLS was far too costly and rigid to provision for larger, more distributed, rapidly changing networks.

SD-WAN promised to address the agility issues but was based on the public Internet. As a result, IT had to build tunnels for every connection. As networks became more complex, building these tunnels destroyed the promised agility gains of SD-WAN.

It is clear that edge computing requires a new, next-generation network.

The next-gen network for edge computing

What IT needs is a new kind of network that provides the performance and security of MPLS, but the agility and cost advantages that SD-WAN promises. What would such a network look like?

  • First, it will have to be private. The public Internet works for consumer-grade applications but falls short for enterprise applications.
  • Second, it will need to provide the efficient, simple routing, performance, and low latency that MPLS provides.

Finally, for edge computing, this next-gen networking must optimize peer-to-peer networks. It must be simple and fast to set up peer-to-peer connections.

An edge computing future

Edge computing completes the journey from fully centralized to fully distributed computing. It opens the door for huge economic and social improvements. But it will also bring the end of the centralized networking era.

As a bonus, the next-gen network required for edge computing will deliver improvements that enterprise computing requires for all aspects of their network: reliable, robust performance, lower latency, better security and privacy, provisioning agility and lowered costs.

About the author

Khalid Raza is CEO and founder of Graphiant. Graphiant is a Silicon Valley venture-funded stealth startup working on next-generation networking technologies.

DISCLAIMER: Guest posts are submitted content. The views expressed in this post are that of the author, and don’t necessarily reflect the views of Edge Industry Review (EdgeIR.com).

Article Topics

 |   |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Featured Edge Computing Company

Edge Ecosystem Videos

Latest News