Method and system for shadow queue based faas scheduler dynamic memory allocation scheduling
By using a shadow queue-based FaaS scheduler to dynamically optimize memory allocation, the problem of low cold start efficiency in FaaS systems is solved, improving system performance and user experience.
CN116795551BActive Publication Date: 2026-07-24WUHAN UNIV
View PDF 0 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUHAN UNIV
- Filing Date
- 2023-06-30
- Publication Date
- 2026-07-24
AI Technical Summary
Technical Problem
The cold start penalty for functions in existing FaaS systems is high, resulting in low execution efficiency. Existing technologies cannot quickly and accurately optimize memory allocation to improve overall performance.
Method used
A FaaS scheduler based on shadow queues is adopted to adjust memory by estimating the gradient of the hot start ratio curve for each function. Combined with the runtime statistics module and the planning and scheduling module, memory allocation is dynamically optimized.
Benefits of technology
This significantly improved the hot start rate of functions, enhancing the overall performance and user experience of the FaaS system.
✦ Generated by Eureka AI based on patent content.
Smart Images

Figure CN116795551B_ABST
Abstract
The application provides a method and system for shadow queue-based FaaS scheduler dynamic memory allocation scheduling, wherein a running statistics module is arranged in the FaaS scheduler, used for respectively estimating the benefits brought by increasing different sizes of memory and the costs required for reducing different sizes of memory for each function; a scheduling planning module is arranged, used for adjusting the maximum benefits brought by different sizes of memory and the minimum costs required for the whole calculation, and deciding the scheme with the maximum net benefits to adjust the memory. The application introduces the idea of the cache problem, estimates the benefit gradient brought by the allocated memory in the form of the shadow queue, and performs the climbing process to realize the optimization of the memory allocation, so that the overall performance of the system can be improved.
Need to check novelty before this filing date? Find Prior Art