Key Criteria for Evaluating Edge Platformsv2.0

1. Summary

Edge platforms leverage distributed infrastructure to deliver content, computing, and security closer to end devices, offloading networks and improving performance. We define edge platforms as the solutions capable of providing end users with millisecond access to processing power, media files, storage, secure connectivity, and related “cloud-like” services from almost anywhere except remote rural locations.

The key benefit of edge platforms is bringing websites, applications, media, and security services closer to fixed and mobile end devices than other types of on-premises or outsourced infrastructure. In GigaOm’s 2021 report “Key Criteria for Evaluation Edge Platforms,” we identified edge platforms coming from four distinct backgrounds:

• Content delivery network (CDN) providers: Owning the distributed infrastructure for content delivery purposes allowed CDNs to enter the edge platform space easily.

• Cloud service providers: Compared to CDN providers, cloud providers had an inverted approach, in which opening more and more data centers has enabled CSPs to create CDN offerings.

• Dedicated edge: Some players have identified the edge platform space as a future market and have quickly built their network from the ground up to deliver these services specifically.

• Telecommunications providers: Typically large organizations with built-in geographical reach and large investment funds, some telecom companies use their in-house expertise and extensive transit networks to expand an edge-focused product portfolio.

The need for content proximity became more apparent in the early 2000s as the web evolved from a read-only service to a read-write experience, and users worldwide began consuming and creating content. Today, this is even more important, as live and on-demand video streaming at very high resolutions cannot be sustained from a single central location. Content delivery networks helped host these media types at the edge, and the associated network optimization methods allowed them to provide these new, more demanding services.

As we moved into the early 2010s, we experienced the rapid cloudification of traditional infrastructure. Roughly speaking, cloud computing takes a server from a user’s office, puts it in a faraway data center, and allows it to be used across the internet. Cloud providers manage the underlying hardware and provide it as a service, allowing users to provision their own virtual infrastructure. There are many operational benefits, but at least one unavoidable downside: the increase in latency. This is especially true in this dawning age of distributed enterprises for which there is not just a single office to optimize. Instead, “the office” is now anywhere and everywhere employees happen to be.

Even so, this centralized, cloud-based compute methodology works very well for most enterprise applications, as long as there is no critical sensitivity to delay. But what about use cases that cannot tolerate latency? Think industrial monitoring and control, real-time machine learning (ML), autonomous vehicles, augmented reality (AR), and gaming. If a cloud data center is a few hundred or even thousands of miles away, the physical limitations of sending an optical or electrical pulse through a cable mean there are no options to lower the latency. The answer to this dilemma is adopting a distributed infrastructure model, also known as an edge platform.