Mass VM Cloning Using Pre-Copied Datastore Disk Pools
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mass cloning of virtual machines (VMs) is time-consuming due to the need to create separate virtual disk copies for each clone, especially in large-scale deployments, and is limited by static resource reservations on host systems.
Innovation Solution
Implement a pre-copying phase to distribute copies of virtual disks across datastores before cloning, creating a disk pool, and utilize these pre-copied disks during the cloning process to reduce the need for new disk copies, allowing for concurrent processing of multiple clone requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate virtual disk copies are created for each clone during mass cloning, then each clone has independent virtual disks, but the cloning process takes a long time to complete
Solution Approach 1:
The patent applies preliminary action by pre-copying virtual disk data from the template to multiple destination datastores before the actual cloning operation. This creates a pool of pre-prepared disk copies that can be quickly allocated to multiple clones simultaneously, eliminating the time-consuming real-time copying process while ensuring each clone receives independent disk copies.
2Productivity
If mass cloning is performed with static resource reservations on host systems, then resource allocation is controlled, but the number of concurrent clone requests is limited
Solution Approach 1:
The patent resolves this contradiction by moving the cloning operation to a different dimension - instead of creating clones directly on host systems with limited resources, the virtual disk data is pre-copied to datastores. This allows multiple clones to be provisioned concurrently from the same datastore pool without contending for host system resources, effectively bypassing static resource reservation limitations while maintaining controlled resource allocation.
3Ease of operation
If virtual disk copies are created in real-time during cloning, then disk copies are available when needed, but the cloning process is slow
Solution Approach 1:
The patent applies preliminary action by pre-copying virtual disk data from the template to multiple destination datastores before the actual cloning operation. This creates a pool of pre-prepared disk copies that can be quickly allocated to multiple clones simultaneously, eliminating the time-consuming real-time copying process while ensuring each clone receives independent disk copies.
Data Source
AI summary
In one set of embodiments, during an initial pre-copying phase, copies of the virtual disks of a VM or VM template can be distributed across one or more datastores, prior to the start of any cloning. This effectively seeds each datastore with a disk pool comprising a certain number of copies of the VM/VM template's virtual disks. Then, at the time of receiving a user request to create k clones from the VM/VM template, up to k copies of the VM/VM template's virtual disks that already reside in the disk pools of one or more target datastores can be moved from those disk pools to the clones' home folders on the target datastores, thereby reducing or eliminating the need to create brand new copies of the virtual disks on demand.