From the course: Serverless and Microservices for AWS

Serverless architectures on AWS - Amazon Web Services (AWS) Tutorial

From the course: Serverless and Microservices for AWS

Start my 1-month free trial

Serverless architectures on AWS

“

- [Instructor] Let's look at a complex, real-world application, and how to approach it with a server-less, microservices solution. Throughout this chapter, I'll discuss a transportation inside dashboard that might be used behind the scenes of a public transit agency in a metropolitan area. In this case, I'll talk about a rail system. But it could also be buses, or ferries, or trams. In the real world, timetables can be published at random times throughout the day and subsequent publications take precedence over previous versions. The system will keep track of where all the trains are located and whether or not they're on time. It will show punctuality metrics in real time and will characterize delays when they occur, so that train controllers can understand if a delay pattern will likely impact other trips on the network. Clearly, there are a lot of moving pieces. Not only on the rails, but also in the system I'll illustrate. Before we begin, we have to look at what data is available to approach the problem. On any given day, there are timetables for each of the various train lines. The transit infrastructure provides the identification and the location of each train once it beings its trip. There is semi-static, geospatial data for the tracks. Calculating punctuality metrics for a particular train requires the relevant timetable and the actual times that the train has arrived at and departed from each station on the route. What are the top level components of the system? There are two primary data pipelines. One for timetables and a second for vehicle position data. First, let's understand the steps that we need to follow to process this data. Call the timetable service if there are new timetables, then download them. Process each timetable package and insert it into our database. The next data pipeline will process vehicle position reports. This data is updated in near real-time for each one of the trains in operation. So, there may be hundreds of trains in operation on the network at any given time. There are few complications that arise that we'll discuss a little later. The steps we'll follow to process the data are: call the vehicle position report service run by the transit agency and serialize the data to JSON for further processing, process the JSON data and insert it into the database. We'll need to determine when each train reaches a station stop. Calculate the metrics. Can we characterize a delay? Cache the results where we can. How we build the delay characterization model in this model, remember, the type of delay matters, so we'll want to know what kind it is.

Contents