Database Resource Competition Detection for Bottleneck Location
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional database management systems face challenges in accurately locating resource bottlenecks due to high concurrency, leading to inefficient resource competition and difficulty in pinpointing performance issues.
Innovation Solution
A database management method that utilizes a competition evaluation model trained on usage features to identify abnormal resources, such as locks or pages, by extracting statistical data and outputting competition alarm information to pinpoint bottlenecks, accompanied by optimization policies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If high concurrency is implemented in the OLTP system, then transaction processing efficiency is improved, but resource competition increases making bottleneck location difficult
Solution Approach 1:
The system implements a feedback mechanism by continuously monitoring resource usage features and feeding them into the competition evaluation model. The model outputs evaluation results that indicate whether bottlenecks exist, creating a closed-loop system that automatically detects and reports bottleneck locations based on real-time feedback from resource usage patterns
Solution Approach 2:
The competition evaluation model acts as an intermediary between the complex high-concurrency system and the bottleneck detection requirement. Instead of directly analyzing raw transaction data, the model receives processed usage features as input and produces interpreted bottleneck evaluation results, simplifying the detection process while maintaining accuracy
2Measurement precision
If statistical collection is performed on database resources, then resource usage information is obtained, but system complexity increases
Solution Approach 1:
The system extracts only the necessary usage features from the complete set of resource statistics. Instead of processing all available statistical data, the competition evaluation model receives a curated set of relevant usage features as input, filtering out unnecessary information and reducing processing complexity while maintaining measurement precision
Solution Approach 2:
The system transforms raw statistical data into standardized usage features that are suitable for model input. By changing the parameter representation from raw statistics to normalized usage features, the system simplifies the data structure and makes it more amenable to analysis without losing important measurement information
Data Source
Figure 1
Figure 2
Figure 3(a)~3(b)
AI summary
A database management method and apparatus are provided. The database management method includes: obtaining statistical data of first database resources in a database, where the statistical data is usage data of the first database resources that is obtained through statistical collection; extracting a first usage feature of the first database resources from the statistical data; inputting the first usage feature into a competition evaluation model, to obtain an evaluation result, where the competition evaluation model is obtained through training based on a second usage feature of at least one database resource; and outputting competition alarm information when the evaluation result indicates that a second database resource is an abnormal resource, where the competition alarm information indicates that a competitive bottleneck occurs in the second database resource, and the second database resource is some or all of the first database resources.