Backup Management System Adaptive Mode Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data backup methods struggle to predict the elapsed time of incremental backups effectively, leading to inefficient backups when changed blocks are scattered over a storage device with small granularity, often taking longer than full backups.
Innovation Solution
A time series model is developed to predict backup time by determining historical times for addressing and copy operations, allowing for adaptive switching between incremental and full backups based on predicted times, optimizing backup efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If incremental backup is performed on scattered changed blocks, then backup data amount is reduced, but backup time increases
Solution Approach 1:
The backup process is segmented into two distinct phases: a prediction phase that estimates backup time before execution, and an execution phase that performs the actual backup. This segmentation allows the system to predict whether incremental backup will be time-consuming and switch to full backup proactively, resolving the contradiction between reduced data amount and increased time consumption.
Solution Approach 2:
The system performs preliminary prediction of backup time using historical data and statistical models before executing the incremental backup. This preliminary action enables proactive decision-making to switch to full backup if prediction indicates time consumption will exceed thresholds, preventing the time loss issue before it occurs.
2Loss of time
If full backup is performed instead of incremental backup, then backup time is reduced, but backup data amount increases
Solution Approach 1:
The system dynamically changes the backup mode parameter (incremental vs. full) based on predicted backup time and historical performance data. By monitoring parameters such as prediction accuracy, actual vs. predicted time ratios, and storage device characteristics, the system adapts the backup strategy to optimize the balance between backup time and data amount transferred.
Solution Approach 2:
The backup strategy is made dynamic through continuous learning from historical backup data. The system updates prediction models based on actual backup performance, adjusting future predictions and decisions. This dynamic adaptation allows the system to optimize the trade-off between backup time and data amount based on changing storage device performance and data characteristics.
3Productivity
If backup mode switching is implemented, then backup efficiency is improved, but system complexity increases
Solution Approach 1:
The system implements feedback mechanisms where actual backup performance data is collected and used to update prediction models. The feedback loop compares predicted vs. actual backup times, refines prediction accuracy, and improves future mode switching decisions. This feedback-driven approach automates the complexity management, making the system self-optimizing rather than manually complex.
Solution Approach 2:
The backup system performs self-service through automated prediction and mode selection without requiring manual intervention. The prediction model automatically analyzes historical data, estimates backup times, and decides whether to switch between incremental and full backup modes. This self-service capability handles the system complexity internally while presenting a simple interface to users.
Data Source
AI summary
Embodiments of the present disclosure relate to a method, apparatus and computer program product for managing a data backup. The method comprises determining a first data amount to be involved in an addressing operation and a second data amount to be involved in a copy operation for an extent to be backed up on a source storage device, the addressing operation addressing a starting address of the extent and the copy operation copying an amount of data corresponding to a length of the extent. The method further comprises obtaining, based on an identifier of the source storage device, a first historical time elapsed for a previous addressing operation having the first data amount and a second historical time elapsed for a previous copy operation having the second data amount.