一种计算任务的调度执行方法、装置和可读存储介质
By introducing an execution plan generator and a resource scheduler into the big data computing system, a horizontally hierarchical execution plan is generated, which solves the problem of uneven resource utilization in the MPP and BSP models, realizes efficient utilization of computing resources, and improves resource utilization and computing efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU BAITWACK TECHNOLOGY CO LTD
- Filing Date
- 2025-07-14
- Publication Date
- 2026-07-17
AI Technical Summary
In existing computing models for big data and data warehousing, there is room for improvement in resource utilization, especially in MPP and BSP models, where uneven resource utilization of worker nodes leads to differences in resource utilization within time slices.
By introducing an execution plan generator and a resource scheduler into the computing system, a horizontally layered execution plan is generated. Resources are evaluated and configured according to the type of operator. By utilizing the operator-level services of the cloud computing platform, on-demand and delayed allocation of operator resources is achieved, thus solving the barrier synchronization waiting problem caused by data dependencies between upstream and downstream operators.
It significantly improves the average utilization rate of computing resources, reaching or exceeding 80%, thereby enhancing resource utilization efficiency. Furthermore, through fine-grained operator scheduling capabilities, it reduces computing costs and latency.
Smart Images
Figure CN120803661B_ABST