Edge data centers are rapidly gaining popularity for a simple reason: they deliver faster services with minimal latency.
Edge data centers facilities, positioned close to the customers they serve, are designed to efficiently deliver cloud computing resources and cached content to end users. The facilities typically connect to a larger central data center or multiple data centers. By processing data and services as close as possible to end users, edge computing allows organizations to reduce latency and improve overall performance.
For IT leaders looking to deliver both resilience and performance, edge computing provided by edge data centers has the potential to be a transformative technology. IT market research firm IDC predicts that the global edge computing market will reach $250 billion by 2024, a compounded annual growth of 12.5 percent. Gartner, meanwhile, forecasts that by 2025 approximately 75 percent of enterprise-generated data will be created and processed outside of the traditional data center or cloud.
Edge data centers are valued for their ability to perform local data gathering and processing while maintaining a high level of availability. Organizations that design application and business processes to run in an independent manner can have survivability of critical business functions, said Carl Fugate, cloud and edge network center of excellence lead at business and IT consulting firm Capgemini Americas. "When designed properly, this can mean business as usual or maybe slightly reduced business functionality in the event of simple failures, like the loss of WAN connectivity," he noted. "There are also key benefits for IoT where the loss of data, or the inability to process data and respond in real time, can render systems ineffective or unavailable."
Edge data centers can deliver performance and efficiency benefits to offices, teams, retail locations, and other widely distributed sites. "Edge data centers remove much of the complexity that comes with forcing all offices, locations, and workers to go through a non-edge centralized data center," said David Linthicum, chief cloud strategy officer at Deloitte Consulting.
Edge data centers also appeal to organizations with business or technology functions that can't rely on conventional WAN data connections, as well as entities that require real-time processing and storage of locally generated data. "We see this in manufacturing, distribution, energy, hospitality, and retail, where services can't be fully reliant on centralized services," Fugate said. "We also frequently see this in environments with operational technology (OT) networks, where sites have machines and processes that are interdependent locally."
IT organizations considering a move to edge computing should begin their journey by inventorying their applications and infrastructure. It's also a good idea to assess current and future user requirements, focusing on where data is created and what actions need to be performed on that data. "Generally speaking, the more susceptible data is to latency, bandwidth, or security issues, the more likely the business is to benefit from edge capabilities," said Vipin Jain, CTO of edge computing startup Pensando. “Focus on a small number of pilot projects and partner with integrators/ISVs with experience in similar deployments."
Fugate recommended examining business functions and processes and linking them to the application and infrastructure services they depend on. "This will ensure that there isn’t one key centralized service that could stop critical business functions," he said. "The idea is to determine what functions must survive regardless of an infrastructure or connectivity failure."
Fugate also advised determining how to effectively manage and secure distributed edge platforms. "The consolidation of services to [the] cloud has made management and security easier through the tools offered by cloud platforms that may not be available with some edge deployments," he observed. "It's important to take this into consideration, as things such as patching, backups, and security can be much harder to implement and manage at remote sites."
The key considerations in planning a highly federated data center model are automation, security, and resilience, said Simon Pincus, vice president of engineering at network monitoring company Opengear. "Recent high-profile failures of content delivery networks (CDNs) have shown how much damage can be done to a business if a service isn't reliable," he noted. "From the first planning session, organizations should consider failure scenarios and how networks and services will be managed and restored."
Pincus also suggested that edge network designers should consider separating the management plane from their primary network to allow operations to continue even when primary connectivity is lost across a distributed network. "Ideally, the management plane will support network automation to provide reliable deployment, reconfiguration, and monitoring," he said.
The diversity and number of platforms available for edge deployments allow for a great deal of flexibility. "In order to ensure the broadest potential for your organization, focus on developing teams and partnerships capable of building cloud-independent solutions," Jain recommended. "There are a number of common architectures and open-source technologies available that you can use to increase the pace of development and also avoid vendor lock-in."