Succeeding at the Edge

Edge computing brings new rules of engagement for digital services. It has a direct impact on competitive market dynamics, creates opportunities for new entrants and enables new services that cannot be supported by conventional cloud architectures. Edge implementation strategies are diverse and greatly influenced by incumbent solution providers as they vie for market superiority. For example, cloud service providers are providing edge computing solutions that are anchored to their cloud platforms. Meanwhile, industrial solution providers have proprietary edge platforms that target the vertical markets they serve, and embedded solution providers are augmenting their existing end-point solutions with cloud connectivity. In contrast, communications service providers (CSPs) generally prioritize edge compute architectures that align with their network points-of-presence.

The diverse approaches for edge network deployments have resulted in numerous standardization initiatives and implementation frameworks. Some, like Multi-Access Edge Compute (MEC) and Central Office Re-Architected (CORD), are being spearheaded by the CSP community. Others include the Fog Consortium, the Edge-X Foundry, the Open Edge Computing initiative and many more that target specific markets. This creates tremendous fragmentation, which is a common phenomenon for nascent markets, but must be addressed before edge computing can reach its full market potential.

In the past, CSPs saw value move away from their networks towards the ecosystem peripheries, benefitting cloud service providers and device manufacturers. If CSPs play their cards right and maintain competitive positioning against web-scale providers, they are well-positioned to use edge computing to bring value back to the network. However, until the edge computing market matures, it remains unclear exactly how CSPs should play their cards. Today, the edge computing strategies pursued by CSPs are nuanced by competitive dynamics, early market bets and legacy infrastructure. For many CSPs, these strategies consist of two-pronged approaches: one for using edge computing to support their own network virtualization efforts and the other to deliver end-user services. While CSPs favor collocated edge computing platforms for both virtualized network and end-user services, they recognize that the dominant use cases will dictate where the edge computing infrastructure resides. Furthermore, since edge computing platforms can be expensive, CSPs are generally taking measured approaches toward the number of edge compute nodes implemented in their networks, paying careful attention to the implied economics. For example, a Tier 1 CSP in the U.S. recently stated its plans to implement hundreds, as opposed to thousands, of edge compute nodes in central office locations to support its current core and radio network virtualization strategy. With this implementation, the CSP stressed the importance of automated management and orchestration solutions and other automated tools to ensure that operational costs and complexities don’t become unwieldy.

Although edge computing has many use cases, the associated business cases will dictate how the use cases are supported, if at all. For example:

• Ubiquitous, ultra-low latency might be achievable in dense network environments but is likely prohibitively expensive for CSPs with large geographical footprints because of the significant number of edge compute nodes needed. Instead of being implemented ubiquitously, this functionality is more likely to be implemented as ‘coverage islands’ to address specific areas where the capabilities are needed. A similar approach might be needed for CSPs when they virtualize their access network equipment, such as cloud radio (cRAN).
• Edge compute solutions must account for the placement of key service and application functionality. If key functionality, such as media content in the case of low-latency entertainment services, cannot be deployed and optimized for edge operations, then the objective of the service will not be met. This was exemplified in a recent demonstration by a Tier 1 CSP in the Asia Pacific region, as an edge application for a media service was found to have a response time of two seconds regardless of the network latency that the edge computing solution enabled.

Edge computing is still nascent but is already enabling innovative solutions and creating opportunities for CSPs, cable MSOs and other ecosystem players to disrupt traditional cloud-based digital service ecosystems. Edge computing will complement rather than replace centralized cloud solutions. The combination of centralized cloud and edge computing will propel digital services into areas that have been unattainable in the past and create new use cases and business models that have yet to be conceived.