Storage Replication Load Prediction for Volume Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems face performance bottlenecks when deploying new secondary and journal volumes in sub-sites due to unpredictable load increases, which can degrade system performance.
Innovation Solution
A management program predicts the load on nodes in the sub-site for new secondary and journal volumes, considering both redundancy processing and non-redundancy conditions, to accurately select optimal deployment destinations based on load predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If new secondary volume and journal volume are deployed in the sub-site, then storage capacity and redundancy are improved, but system performance degrades due to unpredictable load increases
Solution Approach 1:
The management program performs preliminary load prediction before volume deployment by calculating expected processor loads under both redundancy and non-redundancy conditions. This advance analysis allows the system to identify suitable deployment destinations that will not cause performance degradation, thus preventing the contradiction between improving redundancy and maintaining performance.
Solution Approach 2:
The system implements a feedback mechanism where load prediction results are used to guide deployment decisions. The management program continuously monitors node loads, predicts the impact of new volume deployment, and adjusts deployment choices accordingly. This closed-loop control ensures that redundancy is improved while system performance is maintained through informed decision-making.
2Reliability
If redundancy processing is executed for the new secondary volume, then data reliability is improved, but processor load increases
Solution Approach 1:
The system applies local quality by differentiating the treatment of volumes based on their redundancy requirements. The management program calculates separate load predictions for scenarios with and without redundancy processing, allowing selective application of redundancy only where necessary. This targeted approach improves data reliability for critical volumes while minimizing unnecessary processor load increases.
3Measurement precision
If load prediction is performed for both redundancy and non-redundancy conditions, then deployment accuracy is improved, but analysis complexity increases
Solution Approach 1:
The management program segments the load prediction analysis into distinct scenarios: one with redundancy processing and one without. By dividing the complex prediction task into manageable segments, the system can accurately compare different deployment conditions and select the optimal destination. This segmentation improves deployment accuracy while keeping each individual prediction calculation relatively simple and efficient.
Data Source
AI summary
The influence of new deployment of volumes on the performance is predicted with accuracy in a storage system that includes a main site holding a primary volume and a sub-site holding a secondary volume and a journal volume. The main site has one or multiple nodes. The sub-site has multiple nodes. In creating a new secondary volume and a new journal volume in the sub-site, a management program that operates in any node or a predetermined management apparatus predicts the load of the processor under an operation condition in which redundancy processing is executed in the sub-site regarding the new secondary volume, predicts the load under an operation condition in which redundancy processing is not executed in the sub-site regarding the new journal volume, and selects a node that becomes the deployment destination of the new volumes on the basis of the result of the prediction of the load.