This topic describes the common scenarios of Function Compute, including web applications, data extract, transform, and load (ETL) processing, AI inference, and video transcoding.
Web applications
You can use Function Compute together with other Alibaba Cloud services to quickly build scalable web applications. Engineers need to only write business code to build applications. The applications can run in multiple data centers to implement high availability. You do not have to manage scalability and backup redundancy.
High efficiency and free O&M: Function Compute performs O&M on clusters, and engineers focus more on the development of business logic. This helps improve the development and O&M efficiency.
Elasticity and high availability: Instances can be automatically scaled and scheduled within milliseconds to handle traffic peaks.
High performance and low cost: Function Compute provides flexible billing methods for a variety of scenarios to save costs.
Smooth migration: Function Compute supports multiple programming languages and custom runtimes, and is also compatible with traditional application frameworks. You can smoothly migrate traditional applications to Function Compute.
Data ETL processing
Function Compute supports a wide range of event sources. You can use the event trigger mechanism that is provided by Function Compute to process data in real time with a few lines of code and simple configurations. For example, you can decompress Object Storage Service (OSS) packages, clean logs or data in databases, and consume Message Service (MNS) messages.
Simple configuration: Function Compute supports a wide range of event source types. You can process event source data with simple configurations.
High flexibility: You can define different processing logic based on different business scenarios.
AI inference
When an AI model is ready to provide inference services after it is trained, you can pack the data model in functions of Function Compute. This way, the code is run only when actual requests are sent.
High efficiency and O&M-free: Function Compute manages the O&M of clusters. AI engineers can focus on the training of algorithm models and the development of business logic. This helps improve development and O&M efficiency.
Elasticity and high availability: Elastic scaling can be performed within milliseconds based on the number of requests, and tens of thousands of vCPUs can be quickly allocated. Computing power is no longer a bottleneck.
Stability and high reliability: Function Compute provides the multi-version feature and supports canary releases for models. This allows you to implement A/B testing on algorithms to reduce risks upon model-launching.
Simplicity and convenience: Function Compute upgrades toolchains to improve the experience of installing third-party libraries such as TensorFlow and PyTorch. You can deploy applications in the cloud with a few clicks.
Video transcoding
The combination of Function Compute and CloudFlows allows you to easily build a serverless video processing system that provides elasticity and high availability. This solution has significant advantages over traditional solutions in terms of performance, cost, and engineering efficiency.
Flexible transcoding: You can configure custom logic in a video transcoding function.
Parallel transcoding: Automatic scaling can be performed in milliseconds based on the number of video files. Multiple video files can be processed at the same time.
Low cost: Function Compute provides flexible billing methods for a variety of scenarios to save costs.
Quick migration: FFmpeg-related commands can be directly migrated to Function Compute, and FFmpeg-based services that are deployed on virtual machines can be easily switched to Function Compute at low costs.
