Lambda architecture

There are new styles of microservices use cases in the context of big data, cognitive computing, bots, and IoT:

The preceding diagram shows a simplified Lambda architecture commonly used in the context of big data, cognitive, and IoTs. As you can see in the diagram, microservices play a critical role in the architecture. The batch layer process data, and store typically in a Hadoop Distributed File System (HDFS) file system. Microservices are written on top of this batch layer process data and build serving layer. Since microservices are independent, when they encounter new demands, it is easy to add those implementations as microservices.

Speed-layer microservices are primarily reactive microservices for stream processing. These microservices accept a stream of data, apply logic, and then respond with another set of events. Similarly, microservices are also used for exposing data services on top of the serving layer.

The following are different variations of the preceding architecture:

• Cognitive computing scenarios, such as integrating an optimization service, forecasting service, intelligent price calculation service, prediction service, offer service, recommendation service, and more, are good candidates for microservices. These are independent stateless computing units that accepts certain data, applies algorithms, and returns the results. These are cognitive computing microservices run on top of either speed layer or batch layer. Platforms such as Algorithmia uses microservices-based architecture.
• Big Data processing services that run on top of big data platforms to provide answer sets is another popular use case. These services connect to the big data platform's read-relevant data, process those records, and provide necessary answers. These services typically run on top of the batch layer. Platforms such as MapR embrace microservices.
• Bots that are conversational in nature use the microservices architecture. Each service is independent and executes one function. This can be treated as either API service on top of the serving layer or stream processing services on top of the speed layer. Bots platforms, such as the Azure bot service, leverages the microservices architecture.
• IoT scenarios such as machine or sensor data stream processing utilize microservices to process data. These kinds of services run on top of the speed layer. Industrial internet platforms such as Predix are based on the microservices philosophy.