Author: Zen and the Art of Computer Programming
Serverless is a new software development model whose main feature is that it only needs to focus on business logic without having to worry about underlying infrastructure-related issues. This deployment model allows developers to focus more on realizing product functions, thereby improving efficiency and reducing operating costs. The serverless architecture provides developers with a serverless environment through various Function Compute or event-driven computing services provided by cloud service providers. Users do not need to worry about server operation and maintenance, configuration and other issues. With the popularity of container technology, microservice architecture and serverless architecture, more and more people are beginning to develop applications based on serverless architecture. In recent years, machine learning (Machine Learning) and deep learning (Deep Learning) technologies have become more and more popular and have been used in more and more applications. Based on these technologies, many companies have gradually begun to try to deploy machine learning and deep learning models on serverless architecture and provide external services through serverless APIs. Currently, there are many machine learning and deep learning models based on serverless architecture on the market, such as TensorFlow Serving, AWS SageMaker, etc. This article will explore how to combine machine learning models with serverless architecture to build a reliable, high-performance, automated application system. This article will explain in detail how to integrate machine learning and serverless into the serverless architecture by showing you AWS's SageMaker and Tencent Cloud's SCF.
2. Explanation of basic concepts and terms
Serverless architecture generally consists of four layers: infrastructure layer, application layer, function layer, and event layer. Among them, the infrastructure layer includes resources such as cloud platform, storage, network, logs, and monitoring. The application layer consists of a series of functions, each function represents an independent functional unit, and they are connected together through triggers or API gateways. The function layer is responsible for running the actual business logic code, which can be triggered in an event-driven manner or executed regularly. event level rules