Red Hat’s Jaromir Coufal explains the three major adoption patterns emerging in edge computing and offers advice for implementing it within your company.
Edge computing is being deployed in virtually every industry, fueled by the collection of Internet of Things (IoT) data and driven by a vast array of business cases that edge computing can be applied to. However, companies still struggle with scaling out, managing, securing, and automating edge and IoT processes.
3 adoption patterns in edge computing
“What we know now is that there are three major adoption patterns that have emerged for edge computing,” said Jaromir Coufal, principal product manager at Red Hat.
“The first is deploying edge computing at remote offices, such as in a distribution of Walmart or Starbucks stores. The second example is telcos deploying radio networks with multiple tiers of infrastructure for mobile computing. The third is the use of IoT and other devices at the edges of enterprises, such as in manufacturing plants equipped with sensors.”
Regardless of the style of deployment, businesses across the board are experiencing challenges as they place more computing at the edge. One of these challenges is scale.
“When you have hundreds or thousands of locations, it’s a challenge to manage all of that compute as you continue to scale it out at the edge,” said Coufal. “For organizations heavily involved with IoT, there are cases where these enterprises can find themselves with millions of different endpoints to manage. This is where you need to automate as much as you can operationally so there is less need for humans to manage the day-to-day activities.”
To implement the automation, you need experts to set the rules that the automation performs and to monitor the automation.
In the era of more traditional data centers, these infrastructure management tasks often fell to system programmers; however, today, given the need to scale out networks to control all of the endpoints that organizations are deploying at the edge, there are new and emerging needs for skills in networking.
“Networking skills are important at the edge because you need highly skilled people who can make the decisions, such as whether they want to deploy one large network or a series of smaller, specialized networks,” said Coufal. “These same network architects need to make decisions about which of their different networks under management should be federated with each other for information exchange and which they want to keep separate. In many cases, business security and information exchange requirements will dictate this.”
Tips on deploying edge computing
Coufal recommends that organizations take a measured approach when it comes to deploying computing at the edges of their enterprises.
“This means pushing out portions of applications to the edges of your company, but not necessarily everything,” he said. “You can always plan to scale out later.”
It’s also important to place an emphasis on the security that will be needed at the edge, given that end user personnel, not necessarily IT, will be running and maintaining much of this edge computing.
Finally, bandwidth is an issue. If you can place subsets of your data and your applications at the edge, the processing of data, as well as the data that is transmitted from point to point, will be faster.
“There is no single recipe for edge computing that fits every company,” said Coufal. “But a good rule of thumb is that you centralize your computing where you can and you distribute it where you must. You reduce your risk and the latency of your data the more that you can centralize it, so you want to thoroughly assess your use case and its benefits to confirm that an edge deployment is best. In other words, don’t just do edge for edge.”
Big Data Insights Newsletter
Master the fundamentals of big data analytics by following these expert tips, and by reading insights about data science innovations.