What is “the Edge”?
Mobile Edge Computing (MEC) or Multi-Access Edge Computing, is defined in many ways. In fact, the definition of MEC varies widely by the context you consider it, and sometimes by audience. In this post we will explain the nomenclature and concepts that define telecommunications service providers’ network edge and its use in the delivery of mobile, business and residential services. First let’s take a look at the broader term edge. What is it?
The edge, in the traditional usage, has referred to the point where a “customer connects to the provider.” The provider being the organization providing a service. Largely, this was one of three situations:
- An enterprise customer connecting to a service provider’s (SP) edge for network services.
- A retail customer connecting to mobility services.
- A home user connecting to broadband services.
As we know, today we live in a world focused on cloud service providers (CSP). CSP’s are not primarily concerned with network services, but rather providing a place to easily run various workloads at scale. This includes compute, storage, network, AI/ML, IoT, databases, etc. So what is the “edge” in this context?
This brings us to the latest usage of the term “edge.” In this context, the edge is focused on “where the workload is located.” Do you see what just happened? Two things changed. First, we made a pivot from network centric services to workload centric services.
Considering a portion of many organization’s workloads have migrated to CSPs, it’s not surprising focus has shifted away from network to application workloads. Secondly, we are seeing the industry exploring the geography or proximity of the workloads to its users. There may be significant advantages when workloads are placed closer to the user or where the data is generated (more examples to come). The diagram here illustrates the current definition:
Here is the newer definition of edge given the context above:
The edge is a new model of bringing technology resources, including compute and related infrastructure, closer to its end-consumer in order to increase performance and expand technical capabilities.
What is “MEC”?
Now that we’ve taken a closer look at the edge, let’s look at a special use case for service providers. Just when you thought all the cool leading edge toys were only for the cloud companies, we’re right back to talking about service providers again.
Today, for example, when a 4G connected device attaches to a SPs mobile network, most of the mobile applications or “the core,” also called Evolved Packet Core (EPC), are centrally located in large mobile data centers, and therefore further away from the end-user. When a SP moves mobile workloads closer to the user, to increase throughput and reduce latency, we end up with a new mobile architecture called MEC.
According to ETSI, MEC is defined as follows:
“MEC or Mobile edge Computing provides an IT service environment and cloud-computing capabilities at the edge of the mobile network, within the Radio Access Network (RAN) and in close proximity to mobile subscribers.”
Depending on where the MEC nodes are placed or located, there are two flavors of MEC. The first case, is when the MEC node resides “inside” the service provider’s domain. The second, is then the MEC node resides on the customer’s premise.
- Telco Mobile Core Network: This is where the MEC concept was originally established. This use case is known as Mobile Edge Computing or Multi-access Edge Computing (MEC). This span use cases from CS/Passive Optical Network (PON)/Mobile Telephone Switching Office (MTSO), virtual central offices (VCO), Video Hub Office (VHO), Soft Handover (SHO) and others.
- Telco Customer-Premises Service Providers: This concepts points to an architectural design with smart edge device providing the traditional terminations (i.e. NTU, NTD, ONT, STB’s, etc.). At the same time, these new edge devices are capable of running value added or 3rd party specialized services. Example of these 3rd parties specialized services can be: cloud native workloads, IoT gateway services or even extend CSP’s services to customer’s premises.
What is “fog computing”?
Edge computing is considered to be a subset of another overarching concept: “fog computing.” The truth is that fog computing is the superset that defines the functional characteristics of edge computing, but more importantly, it works to improve the efficiency of the data transported to and from the edge device to the cloud (private or public) for further processing, analysis and long term retention. Because of this, many terms and concepts across fog computing, edge computing and MEC are used interchangeably.
We can visualize the relationship among these terms with this conceptual diagram.
What does edge computing look like for different use cases? Here are three example use cases for the edge in the enterprise.
- Enterprise/Consumer Edge: This concept points to an architectural design where the edge device provides the network termination functions while capable of delivering enterprise NFV, on-premise CSP’s functionalities, IoT services, software-defined wide area network (SD-WAN) like universal Customer Premises Equipment (uCPE), or even capabilities for the organization to deploy their remote workloads as cloud native workloads. This includes domain specific specialized features for retail, healthcare, financial and others.
- Oil & Gas/Mining Edge: The concept behind this edge design tends to be related to use cases combining IoT sensors data and actionable results. In these scenarios, some BigData analytics and certain AI functionalities need to happen in real-time or near real-time at the remote location. Examples of these remote locations are oil rigs, oil platforms, or equipments in remote mines. In these setups, network interruptions and edge isolation are expected, but the workloads must be executed to determine immediate actions. Eventually, certain or all data processed by these edge nodes is synced back to the organization’s private or public cloud services.
- Aerospace and Defense Industry Edge: For these particular industries, the edge use cases span across the enterprise & Telcos use cases. Some unique characteristic of this industry is the fact that an edge can be disconnected for months or years (think submarines or autonomous satellites) before being able to send their data back to the organization’s cloud services. Even with these extremes, these edge (which could be multi-node clusters) follow common edge computing characteristics.
As we can see, the functionalities of an edge varies among use cases. Even with these multiple scenarios, at the end, edge computing has a common sets of characteristics, attributes and capabilities. The use case specific functionalities are developed on top of a common set of characteristics and attributes. This is what makes it imperative to revisit the assumptions developed in the past, and consider modern cloud native friendly designs.