Whenever service providers talk to their customers about edge computing, there is always one pertinent question that every customer asks – what is their edge? And how does the service provider define it? The answer to this question is that edge computing can be defined based on the location where the system is deployed and the capabilities that it possesses.
Let us consider the first reference point, which is the location of these computing resources. If we consider this element, then we are talking about some of the standard edge types such as sensor edge, device edge, router edge, branch edge, local area network edge, enterprise edge, data center edge, cloud edge, and mobile edge.
However, instead of absolute distance, if we look at a relative distance, then we can look at some other parameters such as near-edge or far-edge. If the reference points are defined based on capabilities, then they will be defined as thin-edge, thick-edge, micro-edge, and intelligent-edge.
(Also Read: What is Edge Computing? – All You Need to Know)
Types and examples of Edge Computing
Now we will divert our attention to the different types of edge computing and some of the examples based only on the physical location of computing resources.
1. Sensor Edge
If we look at a typical closed-loop system, then sensors are the initial point that sends out events to the backend systems. For instance, if we look at the functioning of a video camera, then the optimal method is to send out live streams whenever there is a motion.
Motion detection and tripwire detection are some of the capabilities that can eliminate the process of sending constant and continuous traffic to the cloud server. These functionalities require edge computing at the sensor level. In most cases, edge computing utilization is very minimal in this case.
2. Device Edge
There are different types of devices that are deployed by customers to execute specific types of functions. For instance, we are talking about specific devices such as X-ray machines, vending machines, motors, and so on. Data can be collected from these devices and analyzed so that it can help in the seamless functioning of these devices. In this case, the computing resources are deployed in closer proximity to the devices so that the data processing of workloads can be done easily.
3. Router Edge
If we look at the primary function of a router it is to deliver packets between networks. They are essentially the differentiating factor between the internal networks and external systems. There are few enterprise routers that provide in-built compute modules and can be used to host applications.
4. Branch Edge
A branch is defined as an office that is different from the head office and is created to perform a specific type of function. Every branch does use different types of applications depending upon their requirement the role that the said branch plays. For instance, in the retail sector, this system could be a point-of-sale application that is used at the storefront, or if we talk about a medical center, then we are talking about Electronic Medical Records.
These are business-critical applications, and they require to be hosted on an edge network at the branch level to ensure that the applications don’t have any sort of latency while users are accessing these systems.
5. Enterprise Edge
When organizations operate in a distributed environment, with multiple branches across a location, the computing resources can be used by these branches in a shared mode. This is primarily to achieve economies of scale and ease out the process of management.
In this method, instead of having edge computing devices installed at each location, it is hosted on a shared site, which is connected to the enterprise network.
6. Datacenter Edge
As we see today more customers are shifting to the cloud network from their current data center. Due to this, we are seeing small data centers are sprouting up so that rapid deployment and data portability on specific events can be done easily in this case, the edge can be deployed closer to the customer.
7. Cloud Edge
Cloud service providers deliver specific services that are closer to the customer. This is to ensure that functions such as content delivery are working in an optimized manner. There is sometimes a reference of cloud edge to Content Delivery Networks (CDN), but these were not developed to host general workloads.
8. Mobile Edge
All the wireless service providers deliver the services in a distributed network. The service locations are relatively closer compared to cloud or datacenter edge computing. When we combine the objectives of these multi-purpose locations, then this model in itself is a very unique one and delivers some key benefits. In this edge computing model, the computing resources are deployed on service access points (also known as SAP).
These service access point locations are based at the core and the applications that operate on these edge computing servers can be accessed through various mobile endpoints by using 4G or 5G network connectivity.
Other than the location, if we consider latency as another parameter for evaluating mobile edge computing, then there are 5Cs of latency that influences mobile edge computing. These are:
- Connection
- Capacity
- Closeness
- Contention
- Consistency
While other edge computing methods may have some advantage over the mobile edge, they will only be in some specific parameters. However, when we take a holistic view then mobile edge does deliver the right balance. In most of the other models, the hardware component is located at the customer site.
This requires additional efforts to handle the space, cooling, power, and physical safety of the hardware component. On the other hand, mobile edge computing ensures that users can consume all the applications as services. This makes it easier for customers to access applications that have low latency, without any hardware deployment in the network.
Final Thoughts
To summarize, every edge computing model does have strengths and its own share of challenges. Experts usually recommend starting with the requirement of customer’s applications and then proceeding with evaluation and selection of the right and best edge computing model.
