Device management at the edge

Edge computing presents a new paradigm for managing computing devices. By placing services closer to the user or the source of data, users get faster, more reliable services, while companies benefit from the flexibility of hybrid cloud computing. When an organization introduces complexity into their device ecosystem, however, they need an additional layer to help them centralize the management and orchestration of edge infrastructure and applications. 

Announced today, Red Hat Edge Manager is a full lifecycle fleet management solution for edge devices at scale, streamlining edge computing infrastructure and application management with end-to-end lifecycle management from security-hardened onboarding to decommissioning, including OS updates, configuration updates and workload management. In addition, it provides registration, provisioning, deployment, monitoring and operations maintenance. 

In this first installment of a 3-part blog series, you’ll be introduced to the concepts behind fleet management at the edge, the problems fleet management solves and how using Red Hat Edge Manager can benefit you and your organization.

What is fleet management?

Fleet management is a very broad concept that means different things to different sectors.  At its core, a fleet is defined as a group of devices managed by a common edge device template and policies. Fleet management originally started in the transportation sector in the 1960s, when trucks in the United States were carrying goods across the country. Truck companies needed a way to manage “fleets” of cargo, but the technology wasn’t advanced enough to accomplish this at the time. As GPS and mobile technology came on the market, companies were then able to solve some of the problems with coordinating their fleets.

Fleet management expanded with the introduction of the Internet of Things (IoT), where companies could coordinate smart devices connected to a network and transmit data between other devices. Today, fleet-level management allows users to define a template and manage policy for a fleet of devices that the system automatically applies to all current and future member devices.

In the context of edge computing, fleet management addresses a number of concepts:

• Coordination and collaboration: Given the challenges at the edge (i.e., locations, connectivity, scale, personnel), users need a centralized way to deploy and manage a fleet of thousands, sometimes millions, of devices.
• Resource allocation: Depending on the configurations and requirements of a given system, compute resources such as CPUs and GPUs may need to be applied differently for different types of workloads. For example, data-intensive AI workloads demand high-performance GPU compute. It's equally important to optimize system efficiency by intelligently routing each workload to the most suitable compute resource.
• Consistency: Polices and templates should be consistently applied across the fleet of devices. When manually applied, some devices may have different versions than others as updates are pushed out, but a fleet-level manager provides rollout policies that enforce consistent behavior across devices, regardless of their current state. This ensures there is uniformity across deployments and reduces configuration drift—a persistent problem that arises when configurations diverge over time, complicating maintenance and management.
• Monitoring and analytics: Organizations need to be able to monitor devices and conduct analysis on what is and is not running. From a business standpoint, they also need to be able to make predictions and improvements. As part of Day 2 operations, monitoring needs to support the following:
  • An observability agent (OpenTelemetry) that enables customers to capture metrics and logs from devices and send them to the observability stack for further analyses. This allows customers to program based on the resulting data so they can shift resources where they are needed.
  • A global inventory of edge devices/servers and applications.
  • Global/national deployments so an OT administrator can quickly assess the health of edge deployments.
  • Centralized logging, metrics and alerts across all devices, Kubernetes clusters and applications at the edge for ongoing observability.
  • Aggregated data by geographic segments and cluster-level insights to monitor the health of your deployments effectively.
• Types of deployments: To maintain optimal flexibility, customers should be able to use devices in whatever environments they need to. Fleet managers can support this need by deploying in hybrid environments, including on the cloud, multicloud and on-premise.
• Modular architecture: The underlying architecture needs to be flexible and extensible to allow for staged approaches to deployment. Extensibility needs to be supported in 2 ways:
  • Lateral growth into other use cases at the same organization using the same infrastructure, where it makes sense.
  • Evolution of richer use cases over time through visibility into the data journey.

What real-world problems does fleet management solve?

As customers deploy hundreds of thousands of devices at the edge, they need an orchestration and management solution that is able to address the challenges of managing a fleet of edge devices. 

• Personnel support: Edge sites often lack expert IT staff on site and rely on operational technology (OT) staff whose job it is to make sure a system continues to run smoothly. OT staff may not be able to fix issues when they arise, but a fleet manager would allow them to onboard devices (i.e., zero touch provisioning), run updates and troubleshoot to a certain level. In industries like oil and gas, it isn’t feasible to dispatch a technician to troubleshoot devices running on a platform in the open ocean, so fleet management helps minimize the personnel needed to maintain these devices.
• Edge site selection: Many edge locations face connectivity challenges, such as being on private, firewalled networks unreachable from the outside or having low-bandwidth, intermittent connectivity or unreliable network connectivity (e.g., oil rigs in the open sea, drones). Unlike datacenter or cloud-based systems, edge devices may operate in harsh physical conditions and in air-gapped environments, where it would be impractical to rely solely on cloud-based management solutions. Their hardware resources and computing power are also typically constrained, making it more difficult to implement resource-heavy monitoring or real-time analytics. Any solution must offer flexibility, with the option for on-premise management that provides parity or additional value compared to cloud-based alternatives.
• Visibility: Current edge deployment processes are manual and complex. As a result, users may lack information about the location and current status of these devices. Many of these devices use very expensive resources and can be wasteful if left unchecked. Fleet management helps the user have better visibility into the hardware, software security and the whole infrastructure of deployments. They can manage things more efficiently and optimize resources, hardware and infrastructure.
• Edge workload placement: It’s often difficult to assess how to assign resources for applications running on the infrastructure. Resources might be differently allocated for different contexts at the edge.
• Scaling: Edge environments operate at an unprecedented scale, ranging from thousands to millions of devices. Customers need to know how to minimize the time, effort and process required to manage each edge device. They want to continue to grow their large fleets of managed devices with minimal personnel. A fleet-level manager is a centralized way to deploy and manage these devices.
• Compliance and regulations: Edge workloads must handle complex regulatory landscapes. Fleet management can define what compliance needs must be implemented and monitor adherence to those policies. As an example, if a clothing retailer can only sell certain items but not others due to local regulations, fleet management can help regulate and monitor compliance.
• Operations management: Failure within an edge system can be catastrophic, but fleet management helps you monitor your infrastructure. Over time, this process becomes predictive rather than reactive and can help users become more proactive in their operations management. 

Who benefits from fleet management?

Fleet management offers many benefits to software developers and DevOps, architects and security teams and management teams. For developers, fleet management enables them to reuse processes and tools that they are using at the datacenter (i.e., GitOps). For architects and security teams, fleet management can enforce security policies, manage vulnerabilities and provide consistency and workload placement. It’s smart enough to manage deployments regularly and can be enhanced at future dates. For managers, fleet management provides more visibility into failures and helps support personnel and compliance. Additionally, the increase in data helps them be more proactive in operations management and determine return on investment (ROI). 

Red Hat Edge Manager provides these benefits through a simple, scalable platform—built for the edge. You can choose what operating system, host configuration and set of applications you want to run on an individual device or a whole fleet of devices, and Red Hat Edge Manager will designate a device agent that will automatically apply the target configuration and report progress and health status back up.

In the next blog post, we’ll discuss features of the FIDO Device Onboarding process so you can get started deploying Red Hat Edge Manager on your Red Hat Ansible Automation Platform and Red Hat Advanced Cluster Management for Kubernetes deployments. In part 3, you’ll explore a use case for edge management in the retail sector.