During his keynote presentation at ONUG Fall 2020, Paul Savill, Senior Vice-President of Enterprise Services at Lumen Technologies, noted that more companies are recognizing the value of data in three ways:

• Acquire. Corporations are collecting data from sources throughout the enterprise, including at the edge.
• Analyze. Organizations are looking at that data in different ways to produce actionable insights.
• Act. Businesses are applying insights to improve and innovate products and services.

Part of this three-step process involves next-generation applications with specific requirements such as low latency and higher bandwidths.

Applications are no longer sending megabytes of data across the enterprise. They are sending gigabytes of data that require the same low latency of a megabyte transfer. As an example, Paul described a retail-robotic installation that was forced to move from a cloud-based configuration to an on-premise one.

Robotics Use Case

The retailer deployed robots to assess a location’s inventory. Information was continuously being sent to a central location in the cloud to improve inventory management. Unfortunately, the network structure degraded performance, and the applications were moved to on-premise. For the retailer with hundreds of locations, on-premise licensing makes the solution less attractive as the cost increases per instance.

Because of the cost savings that can result, many companies are moving applications to the cloud. However, a number of businesses are keeping applications on-premise because of performance concerns. They are not concerned with up-time performance as much as they are about latency. Businesses need low latency when connecting to multiple locations such as branch offices, store locations, or factory floors to ensure operational effectiveness. 

Edge Alternative

An alternative to the on-premise deployment would be an edge computing platform. In this configuration, the edge platform is located off-premise at a distance that is central to other retail locations, but it also maintains the low latency that the application requires. In this configuration, the on-premise devices communicate with the edge computing data center where the applications reside.

At the edge, the applications interface with devices over a dedicated connection. On-premise devices usually have limited functionality but require high performance as the applications interact with the devices. The edge platform also requires high-performance capabilities to on-premise devices, but it does not require the same performance level when communicating to a cloud-based data center. 

Edge Platforms

Edge platforms provide intermediary locations for applications and data to reside. Although the cloud offers significant cost savings and facilitates a more distributed work environment, latency issues can prevent companies from realizing its full potential. Left with few choices, businesses are forced to support applications on-premise, limiting their ability to deploy advanced technologies such as big data analytics or artificial intelligence.

Retailers want all available space devoted to merchandise. Requiring computing equipment to reside locally eats into that space. It also adds to power costs and requires hardware support. Most businesses lose functionality as applications are moved on-premise because the computing power needed to deliver services takes up too much valuable real estate.

Creating an edge platform that resides between the end-points and the cloud reduces the need for on-premise solutions. Deploying an edge platform at a local site can reduce latency issues because traffic doesn’t travel as far. The platform can also provide additional computational power to allow for more real-time calculations. By forwarding data to the cloud, applications at the edge enable businesses to conduct analysis across an enterprise.

Network Use Case

A final use case involved what Paul termed the “Problem of “The Tubes.” A Seattle cloud provider had a client in Phoenix that was having difficulty with an application’s performance. The latency was too high to allow the application to work effectively. This problem highlighted a situation that occurs when using the public cloud.

The Phoenix company used a local ISP for internet connectivity. The local ISP was routing the traffic to a provider in Los Angeles. The LA company sent the traffic to Atlanta, where Lumen picked up the data and forwarded it to the cloud provider in Seattle. The data from the Phoenix company was traversing the United States twice before arriving in Seattle.

The internet is just a collection of autonomous networks that exchange traffic, which means the precise route is not guaranteed. In the case of the Phoenix company, they needed to implement a solution that allowed for a direct connection between Seattle and Phoenix. With the direct link, the latency issues would be eliminated.

Options

The best option for network connectivity depends on a range of factors, including operational efficiencies and cost. Each implementation should be evaluated to determine which of the following options best meets the requirements:

• Public Internet
• Ethernet
• Fiber
• Wavelength
• IP VPN 

Understanding the application requirements and the deployment configurations can help organizations determine the best way to meet their operational and business needs.

An edge platform provides an alternative that can balance the cost versus the performance equation that many organizations face. Whether it is network functionality, containers, or virtual machines, an intermediary platform can deliver needed functionality more cost-effectively.

What is required for successful implementation are applications that can take advantage of an edge platform’s intermediate functionality. Solutions are required that deliver the functionality needed at the device level without compromising data delivery to system-wide stakeholders. Finding the right configuration also demands an understanding of the underlying network. 

Edge Implementation

Applications can reside at any of the three points in an edge implementation. They can run in the cloud, at the edge, or on-premise. Alternatively, different parts of the application can reside at each location. This deployment requires a tool that can orchestrate the service chain of applications across the enterprise. 

Lumen’s solution was used in a manufacturing facility where robotics were found on the factory floor. Part of the application resided on-premise, but a percentage of the application remained on the edge platform. The data was forwarded to the cloud solution for use across the enterprise. The idea of an edge platform gives application developers another option for how an application collects, transmits, and stores data.

If you want to learn more about network connectivity options, join us at ONUG Spring 2021, May 5-6, where we will continue these discussions, or contact us by clicking here.