What are Edge Computing and Multi Access Edge Computing?

There is some confusion between these two terms - Edge computing and Multi-Access Edge computing.

This is the understanding I went with.

Edge-computing term is used when the compute, storage is brought closer to the users. It is similar to cloud computing, but computing may happen at multiple edges, hence is it used in the context of distributed computing.

MEC (Multi-Access Edge Computing) is also Edge-computing, but with some specific understanding of (access) network context. Access networks include 4G, 5G, Private 5G and possibly others. But, I think many times MEC is referred in the context of public 5G and private 5G.

If a provider says that they are providing MEC, it means following in my view.

• Intelligent placement: Provider is able to place your application in specific 5G tracking areas and Edge locations to meet the performance requirements. Placement constraints include latency, cost, platform capabilities (such as access to accelerators - GPU, Crypto, AI etc..)
• Traffic Prioritization: Provider is able to prioritize traffic through the network to meet latency requirements by translating application requirements to 5G QoS flows, mapping rules etc..
• Intelligent association of client to right instance of distributed application service: Provider shall have ability to associate the client traffic to right instance of application service to meet SLA requirements. There are multiple ways to do this and will talk on this more later.
• A way to provide network context: Provider is able to inform network and user context to applications that are authorized to get this information.
• Application Service Registration and Discovery: Providing a way for service registration and discovery across vendor applications - A way for applications to provide services to others applications and for applications to leverage services from other applications.
• Application context Continuity: Providing a way for sophisticated applications to share the client context from one instance to another when user moves to new tracking area.
• Short Message support: facilities to the applications to send short messages to wakeup devices or wake up client applications.

One hear the terms MEC-aware applications and MEC-unaware applications. MEC-aware applications are the ones that leverage network/user context information to make decisions. Few examples include applications changing compression algorithms based on available bandwidth; applications inserting the different types of advertisements based on the location of the user; applications leveraging external services that are customized for that geographic location. MEC-unaware applications don't have any specific logic in them to take advantage of network context.

One also hear terms such as Enterprise-MEC, Network-MEC and Industrial-MEC. In my view, the terms 'Enterprise', 'Network' and 'Industrial' represent where the MEC being provided. Enterprise-MEC indicates that the provider is able to provide MEC in their customers premises. Network-MEC indicates that the provider is providing MEC in their network locations and Industrial-MEC indicates that the provider is providing MEC at industrial locations. Note that provider can be Enterprise itself, Telco operators and could be dedicated MEC provider. Also, note that in many cases a given Enterprise may require MEC services in multiple places and hence may go for multiple providers.

Some standard bodies are working on to ensure that there are standards for 'Application Service discovery and registration', 'Associating client traffic to the right application instance", "application service continuity". 3GPP SA6 (23.558) is defining elaborate specification to address MEC market. 3GPP SA6 is calling this architecture as 'EDGEAPP'.