Organizations turn to hyper-converged infrastructure to support systems such as relational databases and virtual...
desktop infrastructure, but companies can also use HCI to implement edge computing.
An edge computing platform moves specific data processing workloads to the network's periphery, which brings the data closer to the users who produce and consume it. This reduces latency and improves performance, but compute, storage and network resources must be able to handle the workloads. That's where HCI comes in.
HCI provides an integrated platform to process and store data using minimal resources. For specific edge computing workloads, hyper-convergence offers a number of important benefits, such as ease of maintenance and a reduction of data transmitted back to the primary data center. But it cannot address all the challenges that come with edge computing, such as the tendency toward creating data silos and the management of widely distributed endpoints.
The new age of edge computing
The centralized data center has been the de facto standard for processing and storing data. Cloud computing extended this model beyond organizational boundaries, but the centralized infrastructure remains at the core. Still, mobile computing, distributed workforces and the internet of things (IoT) are pushing IT administrators to reconsider the centralized approach. Some organizations see an edge computing platform as a way to address the bottlenecks and performance issues that come with a centralized model.
For example, internet and WAN service disruptions can severely handicap remote offices, which can degrade application performance and affect employee productivity. The more remote the office, the greater the potential for performance and latency issues.
The influx of IoT data is a different story: The amount of information that IoT devices generate and collect is already unprecedented. There is simply too much data to effectively push all of it to centralized data centers, whether on premises or in the cloud. To complicate matters, many devices require real-time processing and cannot afford the latencies that come with the centralized model.
The processing components an edge computing platform might use can take many forms, depending on the types of applications and data. But the goals are always the same: to streamline traffic flow and maximize performance.
Enter the HCI
Most edge computing operations fall into two broad categories. The first occurs at the point of origin, where an intelligent device has the capacity to process and store its own data, as well as communicate with a central data center.
The second category is the micro data hub located proximally to the users or devices. Possible uses for such a hub include support for a remote office, an intermediary collection point for IoT data or a processing satellite for remote customers. Hyper-convergence can be a good fit for this second category.
Vendors sell HCI appliances with out-of-the-box products that feature quick implementation with relatively few IT resources. Some vendors sell the HCI software as a separate product for organizations that want to assemble their own HCI implementations, often using commodity hardware.
HCI and the edge computing platform
Edge computing comes with a wide range of challenges in planning, configuration, integration, licensing, security and management. An HCI implementation can help address many of these issues, especially a prepackaged HCI appliance. An appliance is easy to deploy, manage and scale, and it comes with minimal configuration and integration concerns. This makes it quick to implement and easy to manage. An HCI appliance optimized for peak performance also requires less power and space than a traditional rack of hardware.
HCI appliances often come with bundled capabilities, such as data compression and deduplication, backup and disaster recovery, as well as automatic failover. IT teams do not have to install and configure these applications, which offers further savings.
Perhaps the bigger issue with hyper-convergence is the risk of assuming that it can address all the concerns that come with edge computing. For example, data and systems management is inherently much more complex with edge computing because workloads are distributed. Although an HCI system can mitigate these issues at the individual, local level, IT must still be able to track and manage the entire corporate infrastructure in a unified manner.
This distribution also creates data silos that are difficult to reconcile across multiple systems. This adds even more administrative overhead. IT admins must address issues such as data redundancy and long-term data retention at a corporate-wide level, something that cannot be resolved solely through the use of hyper-convergence.
One of the biggest concerns with edge computing is security. Edge computing reduces the amount of data that is transmitted across networks, which can lower risks. IT admins must still ensure that each local data center is fully protected from possible risks and that the data is secure at rest and in motion.
Under the right circumstances, HCI can be ideal for implementing an edge computing platform, especially in a location with relatively few IT resources. Embracing such an implementation, however, requires that IT admins have their eyes wide open and that they take the many challenges that come with edge computing into account.