Virtual Data Service Appliance Parallel Disk Image Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current disaster recovery systems face challenges in efficiently testing multiple disk images at different points in time, which increases recovery time objectives and storage costs due to the need for complete data copies, making it impractical to test multiple points in parallel.
Innovation Solution
A dedicated virtual data service appliance (VDSA) within a hypervisor provides data services, including replication and monitoring, by intercepting I/O requests, journaling writes, and replicating data asynchronously across sites, allowing for parallel testing of multiple points in time without duplicating data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple complete disk image copies are created for testing different points in time, then the ability to test multiple recovery points in parallel is improved, but storage consumption increases significantly
Solution Approach 1:
The patent segments the disk image into a base image and multiple incremental updates corresponding to different points in time. Instead of storing complete copies, the system stores the base image once and maintains separate update streams for each point in time, allowing parallel testing without duplicating the entire disk image data.
Solution Approach 2:
The patent implements a nested structure where incremental updates are contained within the framework of the base image. Each point in time representation is nested within the same storage space as the base image, with updates layered on top, enabling multiple temporal versions to coexist without requiring separate complete copies.
2Quantity of substance
If multiple disk images at different points in time are tested sequentially, then storage requirements are reduced, but the recovery time objective increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing and organizing disk image data into base images and incremental updates during normal operations. This preliminary organization allows recovery testing to proceed in parallel by having all necessary components ready and structured for simultaneous access, rather than sequentially processing complete image copies.
Solution Approach 2:
The patent creates lightweight copies in the form of incremental update references rather than complete data copies. Each point in time is represented by a reference to the base image plus its specific updates, enabling parallel testing while maintaining minimal storage overhead compared to full image duplication.
3Reliability
If complete copies of disk images are maintained for each test point, then data integrity for recovery testing is ensured, but the cost and complexity of the system increases
Solution Approach 1:
The patent merges multiple point-in-time representations into a unified storage structure where the base image and incremental updates are combined in a single storage space. This merging approach ensures data integrity across all time points while reducing system complexity compared to managing separate complete copies for each recovery point.
Data Source
AI summary
An enterprise disaster recovery system, including at least one data disk, a processor for running at least one data application that reads data from the at least one data disk and writes data to the at least one data disk over a period of time, a recovery test engine that (i) generates in parallel a plurality of processing stacks corresponding to a respective plurality of previous points in time within the period of time, each stack operative to process a command to read data at a designated address from a designated one of the at least one data disk and return data at the designated address in an image of the designated data disk at the previous point in time corresponding to the stack, and (ii) that generates in parallel a plurality of logs of commands issued by the at least one data application to write data into designated addresses of designated ones of the plurality of data disks, each log corresponding to a respective previous point in time, wherein the plurality of previous points in time within the period of time are specified arbitrarily by a user of the system.