Definitions matter in telecom. With the name change from Mobile Edge Computing to Multi-Access Edge Computing, and now perhaps just Edge Computing, with some use cases that do not seem to involve the wide area network or even the access network at all, does anyone really know what MEC is and does it matter?
At its most basic level, mobile edge computing or multi-access edge computing or just edge computing distributes cloud services compute and storage resources throughout the network to reduce delays and latency, distributes services to support new use cases and revenue opportunities and decreases service deployment costs. MEC does this by merging IT, content, connectivity and location to place compute, storage, orchestration and application resources closer to the users.
The problem comes when you look at where the placement happens and for what applications. Is it a function such as autonomous driving or gaming and virtual reality? Do you architect MEC services on the basis of latency bands? Or on the basis of location such as cell site, pre-aggregation, aggregation, regional and central data centers? The upshot is that MEC means many things to different people, companies and organizations.
The clarity challenge
One of the considerable challenges with understanding edge computing is agreeing to a common definition with your suppliers and partners. Nearly every MEC solution definition starts with "And here is what we mean by the edge." Here is a sample of edge compute definitions by different organizations.
These divergent definitions make what edge is as clear as mud. So, the question remains: Where is the edge? For some organizations such as ETSI/3GPP, HPE and OpenStack it is the edge of the network, which could mean the RAN, the CO or even the CMTS. For others edge means the enterprise location providing the lowest latency with the greatest security and enterprise control.
The architecture conundrum
Another aspect blurring a definition of MEC are the competing issues in defining MEC layers and services: location determines latency, but locations are not the same in every network or even within carrier networks. Aggregation site locations in a dense urban area such as London, Hong Kong or New York City can be several hundred meters or yards apart. In suburban areas distances can grow to miles, and in rural areas may not exist at all. This location issue poses a challenge to ubiquitous services, potentially widening the urban/rural digital divide.
The definition of the edge continues to evolve with some early cell-tower MEC proponents, such as VaporIO and EdgeMicro, now suggesting that cell-site aggregation locations provide a more economical and broader service option for edge compute applications compared to putting MEC resources at the cell tower. This strategy has been echoed by several operators, including BT/EE, SKT, Sprint, Telefonica and Verizon.
Whether edge compute is intended to address application service and performance requirements onsite, at the network edge or deeper into the core, where the edge actually resides is less important than what business problems is it solving.
— Chris Nicoll, Principal Analyst, Wireless & Mobile, ACG