Updated on 2025-07-09 GMT+08:00

Data Plane Assurance

Load Balancing and Network

  • Load balancing for high availability

    Load balancing distributes traffic, preventing single points of failure and improving system reliability.

  • VPC configuration for network isolation

    Computing nodes are in isolated VPCs, strictly separated from external networks for security.

  • Flexible network configuration

    By default, functions have public access, but users can configure access to specific VPC resources.

Scheduling

  • Multi-cluster for enhanced DR capability

    The multi-cluster and multi-AZ architecture allows resource migration during AZ fault, ensuring continuous operations.

  • Intelligent scheduling

    The intelligent algorithm predicts traffic and automatically scales up resources to quickly respond to burst traffic.

  • Elastic and reserved instances

    Function instances are classified into elastic and reserved instances. Elastic instances can be dynamically created based on real-time service load changes and automatically released during off-peak hours to avoid resource waste. Reserved instances are configured and created in advance by users based on service requirements and are not automatically released. You can set the maximum number of elastic and reserved instances based on service requirements.

Function Invocation

  • Synchronous invocation

    Requests are processed directly without caching, suitable for scenarios requiring high real-time performance.

  • Asynchronous invocation

    Requests are cached in message queues to guarantee execution. Queues are isolated by account or function to avoid data interference among users. The system retries failed invocations up to three times by default. And you can customize the retry times.

Runtime Environment

  • Vulnerability fixing and security upgrade

    FunctionGraph regularly scans and fixes vulnerabilities in compute nodes and function instances, ensuring a secure and stable runtime environment.

  • Immutable code

    The code modification takes effect only for newly generated instances and does not affect running instances, ensuring code consistency and stability.

  • Non-persistence environment

    File systems and memory are released together with the instance, preventing data residue and enhancing resource utilization.

  • Exception information collection

    The runtime environment automatically collects exceptions and logs to help quickly identify and resolve issues, improving troubleshooting efficiency.