Dynamic Backup Policy for Data Centers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional backup policies based on Recovery Point Objective (RPO) are inadequate for highly scalable data centers and applications, as they do not effectively measure the amount of data that could be lost in case of a disaster, leading to potential data loss in scale-out environments.
Innovation Solution
Implementing a dynamic backup policy that measures and tracks the amount of data changed since the last backup, using change tracking filter drivers and hypervisor-provided change block tracking to determine when a backup is necessary, allowing for configurable thresholds at various granularities such as file or virtual machine levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional RPO-based backup policies are used, then backup timing is simple to configure, but the ability to protect against data loss in scale-out environments deteriorates
Solution Approach 1:
The backup policy transitions from static RPO-based timing to dynamic change-based triggering. The system continuously monitors data changes and dynamically decides when backup is needed based on actual data state, not fixed time intervals. This resolves the contradiction by maintaining operational simplicity while improving reliability through adaptive, data-driven backup decisions.
Solution Approach 2:
The system implements feedback by monitoring actual data changes and using this information to trigger backup actions. Change tracking mechanisms provide real-time feedback about data modification, allowing the backup system to respond appropriately. This feedback loop ensures reliable data protection while keeping the interface simple for users.
2Device complexity
If static RPO backup policies are implemented, then system complexity is low, but measurement precision of data loss risk deteriorates
Solution Approach 1:
The system replaces mechanical/time-based backup scheduling with a more sophisticated data-driven measurement approach. Instead of relying on fixed RPO time periods, the system uses change tracking mechanisms to precisely measure actual data changes. This substitution maintains reasonable system complexity while dramatically improving measurement precision of data loss risk.
3Reliability
If dynamic change-based backup policies are implemented, then data protection reliability improves, but system complexity increases
Solution Approach 1:
The backup system performs self-service by automatically monitoring data changes and triggering its own backup actions without requiring constant external intervention. Change tracking mechanisms autonomously detect when backup is needed and initiate appropriate backup operations. This self-service capability improves reliability while managing complexity through automation rather than requiring complex manual policy management.
4Productivity
If traditional time-based backup scheduling is used, then production performance impact is minimal, but data protection in scale-out environments deteriorates
Solution Approach 1:
The system uses periodic monitoring of data changes rather than continuous backup operations. Change tracking runs periodically to detect data modifications, and backup is triggered only when changes exceed thresholds. This periodic action maintains production performance by avoiding constant backup interruptions while ensuring comprehensive data protection through continuous monitoring and event-driven backup execution.
Data Source
AI summary
Techniques to protect against data loss are disclosed. In various embodiments, an amount of data that has changed since a last backup of at least a defined subset of a save set is determined. The determined amount of changed data is compared to a corresponding threshold. A backup is performed based at least part on a determination that the determined amount of changed data equals or exceeds the threshold.