From the course: Achieving Low-Latency Data with Edge Computing

Reference architecture

From the course: Achieving Low-Latency Data with Edge Computing

Start my 1-month free trial Buy for my team

Reference architecture

“

- [Instructor] All right. Let's dive into what an edge computing software architecture might look like. Let's refer to the component that does the actual computing as an application or app for short. The app may need a component that handles the read and write of data, which we'll call a data management component. We should also add a component that handles data movement or IO between apps on the edge or between data centers. In this case, the app is dependent on the data management and data movement components. These dependent components are often referred to as backing services. Since both of these services deal with data, I will refer to them as data services. The data management service can be fairly simple. For example, it can read and write data in a straightforward structured format on the disc, such as a text format. In other cases, you may need a more advanced data management component, such as a database. There are different types of databases such as a relational and NoSQL. One of the oldest and most popular database types is relational. A relational database management system, or RDBMS, allows applications to access data through a query language called the structured query language or SQL for short. NoSQL or not only structured query language is a newer database technology that we will be using in our remote patient example. It supports improved scalability and lower data access latency than traditional relational databases. Why you may ask? Well, we'll discuss that during our in memory access and scalability sections. The data movement component does the data IO and it can be a simple module that sends data over a network socket, such as a remote procedure call or RPC based protocol. In some cases you may need a more advanced data movement component that improves scalability and latency such as a message broker. Let's chat more about message brokers in our next module.

Contents