Function Compute is an event-driven, fully managed computing service. You do not need to manage servers or other infrastructure. You only need to write code and upload it. Function Compute prepares computing resources for you and runs your code in an elastic and reliable manner.
Function Compute charges you based on resource usage by default. The resource usage is calculated by multiplying the specifications that you configure for your functions by the duration of use. Resources are allocated only when needed and released promptly. For more information, see Billing overview.
Differences from traditional computing resources
In a traditional client-server model, servers are always running services regardless of whether requests are received.
However, Function Compute follows a serverless architecture, executing functions only when there are incoming requests and promptly releasing function instances when there are not. You pay only for the resources that you actually consume and no longer need to manage servers.
What can Function Compute do
Scenarios | Why use Function Compute | Examples |
Build web applicationsFunction Compute provides ready-to-use templates for popular web applications. You can quickly build and iterate your web applications. As your business expands, you can also use features such as log query, performance monitoring, and alerting to ensure that your web applications run efficiently and reliably. | Function Compute is highly elastic and suitable for web application scenarios with burst traffic, such as sales promotions. | You can use the Flask framework template to create functions and efficiently develop web application code based on the template. You can also migrate existing web applications to web functions. For more information, see Use web functions to quickly create a web application. |
Real-time data processingBased on event-driven architecture, Function Compute can be automatically triggered by HTTP requests, OSS, MSMQ, and other services. You can use OSS to trigger Function Compute to process uploaded files in real time. You can also organize multiple functions, message queues, and databases to collect massive amounts of IoT data. When scenarios change, you can adapt your applications by modifying event triggers and integrating new components without making significant changes to your business code. | Function Compute can integrate with multiple Alibaba Cloud products to easily build event-driven architectures suitable for various data processing scenarios. | You can use Function Compute to automatically decompress ZIP files uploaded to OSS. For more information, see Use Function Compute to automatically decompress ZIP files uploaded to OSS.  |
AI model serviceWhen an AI model is ready to provide inference services after training, you can package the data model in Function Compute functions. This way, the code runs only when actual requests are received. | The GPU instances of Function Compute allow you to focus entirely on your business without managing the underlying GPU infrastructure, greatly simplifying your implementation path. | You can use LLM container images and GPU functions to quickly deploy a chatbot application. For more information, see Deploy Google Gemma model services at low cost based on Function Compute.
|
How to use Function Compute
You can refer to the following quick start tutorials to learn about the operations in the Function Compute console and the development process of Function Compute.
In addition to the console, you can also use Function Compute in the following ways.
Use Function Compute through the Serverless Devs tool. For more information, see What is Serverless Devs.
Use Function Compute through API or SDK. For more information, see SDK Reference.