Service Cluster Workload Prediction Using Cluster-Specific Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing workload prediction solutions for service clusters in cloud computing are limited in their ability to handle diversified online computing service demands and suffer from inaccurate prediction, leading to resource wastage or shortages.
Innovation Solution
A workload prediction method that involves obtaining load indicator sequences, determining service representations, performing clustering to identify category clusters, and using pre-trained sequence prediction models specific to these clusters to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a unified prediction model is used for all services, then the model structure is simple, but prediction accuracy deteriorates and cannot satisfy diversified service demands
Solution Approach 1:
The patent segments the service cluster into multiple category clusters based on service characteristics and workload patterns. Instead of using a single unified prediction model for all services, the system creates separate prediction models for each category cluster, allowing each model to be optimized for its specific service type while maintaining overall system manageability.
Solution Approach 2:
The patent applies local quality by customizing prediction models for different category clusters based on their specific characteristics. Each category cluster receives a prediction model tailored to its unique workload patterns and service requirements, rather than applying a generic model uniformly across all services.
2Ease of operation
If a unified prediction model is used for all services, then the system is easy to operate, but prediction accuracy deteriorates leading to resource wastage or shortages
Solution Approach 1:
The patent segments services into category clusters with dedicated prediction models, which improves resource planning reliability by providing accurate predictions for each service type. The segmentation is implemented in a way that maintains operational ease through automated clustering and model selection processes.
Solution Approach 2:
The system performs self-service through automated service representation generation, clustering, and prediction model selection. The workload prediction system automatically determines which model to use for each service based on its category cluster, reducing manual intervention while improving prediction accuracy for resource planning.
3Device complexity
If existing prediction solutions are used, then the system complexity is low, but prediction precision is limited and cannot accurately predict future workload
Solution Approach 1:
The patent applies preliminary action by pre-processing service data to generate service representations and pre-clustering services into category clusters before making predictions. This preliminary organization of data and services enables more precise predictions while managing system complexity through structured data preparation.
Solution Approach 2:
The patent segments the prediction system into multiple category-specific models rather than using a single complex unified model. This segmentation approach improves prediction precision for each service type while keeping individual model complexity manageable through specialization.
Data Source
AI summary
Embodiments of this specification provide a workload prediction method and apparatus for a service in a service cluster. The method includes: obtaining a load indicator sequence of each service in a service cluster corresponding to a workload indicator in a same historical time period; determining, based on the load indicator sequence corresponding to each service, a service representation corresponding to each service; performing clustering processing based on the service representation corresponding to each service, to obtain a target category cluster to which each service belongs in multiple category clusters; obtaining multiple sequence prediction models pre-trained for multiple tasks, and enabling the multiple category clusters to correspond to the multiple tasks; and inputting at least a load indicator sequence of any service into a target sequence prediction model corresponding to a target category cluster of the service in the multiple sequence prediction models.