ra2 studio - Fotolia
Top guidelines for edge computing software selection
Finding edge software that meets all your organization's needs requires help from both the business and IT sides. Get the right fit with use case information and developer input.
As organizations become more dispersed and data consumption increases, IT departments may find a need to effectively deploy edge devices through software. Edge computing software comes in all shapes and sizes. Enterprises can select the best ones by starting with a few key considerations to narrow down their list of candidates.
First, determine the business case to define the overall needs of an edge project. This informs what edge computing means for the business; there are many possibilities, including application delivery and IoT device support. It's also beneficial to see how the different product options align with existing data center tools for visibility, asset control and developer software deployment.
Consider the business problem
After identifying the business case, an organization must understand the business problem it wants to solve and how fixing the problem fits within the company's existing strategy. This assessment helps determine the potential value of edge computing, and admins can then evaluate different edge computing software that addresses these needs.
Key considerations include raw vendor cost of the potential offerings, cost of implementation, steps needed to integrate the software into operations, security risk evaluation and staff knowledge growth. Organizations should also factor in continuous lifecycle and maintenance costs.
For example, a custom-built configuration may cost more in the long run when future staff and operational costs are considered. But the cost of shifting the company's strategy may have a larger financial benefit. Ultimately, the value of the edge computing setup and software must outweigh all one-time and ongoing costs.
Define what 'edge' means
Part of the selection process is defining what the edge means to the organization. This helps characterize the latency and coverage requirements for the edge software. For instance, enterprises should look at cellphone tower deployments that support less than 1 millisecond latency for certain IoT use cases.
Older edge computing software could tolerate sub-100 millisecond latency for use cases like accelerating software running across the globe. Most enterprise apps can operate with latencies between 200-300 milliseconds, which is sufficient with a data center in the same country.
IoT implementations require that the edge computing software is closely aligned with the operating technologies. In this use case, the software must support much lower latency requirements for applications that involve device health and performance management.
Be sure to evaluate the cost of edge computing software downtime. It might be OK for a video analytics application to go down for a moment, but an autonomous vehicle application failure is much more dangerous.
"Thinking about these things will determine how you should be processing your data," said Chris Hinkle, CTO at TRG Datacenters, a data center co-location provider.
Also test the edge setup's ability to support local data processing. Look for an efficient complex event processor that cleanses, normalizes, contextualizes and aligns raw streaming industrial data.
Some edge computing offerings make it easier to shrink the AI models running on edge devices and reduce bandwidth requirements. Newer edge software also simplifies AI model development in the cloud and automates their deployment.
Use edge computing software for visibility and control
Another important consideration is asset visibility and control. This aspect is important because of the geographic dispersion of edge facilities and the inherent need to incorporate everything into the larger data center strategy.
Key considerations must revolve around remote management, monitoring, support and autonomy. Edge computing setups invariably operate in locations remote from IT. These sites don't have the same compute resources as a data center and aren't guaranteed cloud connectivity.
Visibility can be tricky when the edge software must support legacy code and equipment. Edge deployments are often an extension of existing equipment and applications. Admins should assess whether an edge setup will require new equipment, a switch in cloud providers or code rewrites.
Align with software developers
Be sure to select edge computing software that aligns with the organization's chosen development workflows. This makes it easy to quickly deploy applications without developers having to learn new technologies to take advantage of the software's capabilities.
"It comes down to working with software languages that your teams already know," said Mike Hostetler, senior director of engineering at Cars.com, a digital automotive marketplace.
Admins also want to align edge computing software with a company's continuous integration and continuous deployment tooling.
"If a developer has to deploy one way for data center, one for core cloud and then another for edge, all of that productivity gain will be lost," said Sean Leach, chief architect at Fastly, an edge cloud platform provider.
