Virtual Machine Thin Cloning via Metadata Deduplication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual machine cloning methods are resource-intensive and cumbersome, often duplicating data and requiring significant time, as they involve copying all files associated with a base virtual machine, leading to inefficiencies in managing and deploying multiple virtual machines.
Innovation Solution
The technique involves creating thin clones of virtual machines by taking a snapshot and copying metadata that identifies the physical location of files, allowing for deduplication and efficient deployment of multiple virtual machines without duplicating data, and customizing and registering the clones within a hypervisor environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional VM cloning methods are used to copy all files associated with a base VM, then a complete clone is created, but the process is resource-intensive and time-consuming
Solution Approach 1:
The patent segments the VM cloning process into two distinct phases: (1) creating a metadata reference that points to the base VM's file locations without copying data, and (2) performing actual file copying only when and where differences are detected. This segmentation allows the clone to be created immediately with minimal resources, while maintaining the option to synchronize data later.
Solution Approach 2:
The patent performs preliminary actions by creating the clone's metadata structure and establishing references to the base VM's files before any actual data copying occurs. This preliminary setup enables the clone to become operational immediately, with data synchronization happening asynchronously in the background only when necessary.
2Reliability
If all files are copied to create a VM clone, then the clone is fully independent, but storage space is wasted due to duplication of identical data
Solution Approach 1:
The patent uses a metadata copying approach rather than full data copying. The clone's metadata contains references pointing to the base VM's file locations, allowing the system to access shared data without physically duplicating it. This reduces storage consumption while maintaining the ability to detect and copy only the necessary differences.
Solution Approach 2:
The patent merges the base VM and clone into a unified storage structure where identical files are stored once and shared between both VMs through metadata references. This consolidation eliminates redundant data storage while preserving the functional independence of each VM through their distinct metadata configurations.
3Productivity
If a snapshot is taken to identify file locations for cloning, then cloning efficiency is improved, but additional storage overhead is required for the snapshot
Solution Approach 1:
The patent extracts only the essential metadata information (file location references) from the base VM's file system structure, rather than creating a complete snapshot of all data. This extraction approach provides sufficient information for efficient cloning while minimizing the storage overhead associated with the snapshot process.
Data Source
AI summary
While current solutions for cloning virtual machines can involve copying and duplicating files associated to a virtual machine (VM), systems and techniques can be devised to create thin clones of a VM using the VM's associated storage system to copy and deduplicate storage for the VMs. One can create a base VM in a directory of a storage container attached to a hypervisor, and then map the storage container to a volume on a storage system. One can clone the base VM using a snapshot of respective files associated to the base VM, which can comprise creating metadata that identified a physical storage location of the files. Further, the metadata can then be copied to a desired location on the storage container, to create a VM clone. Once copied, the cloned VM can be customized and registered within the hypervisor.