Consolidating Redo Log Chains in Linked Clone Trees
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Linked cloning in virtualized data centers experiences performance degradation due to increasing chain lengths of redo logs and branching disk trees, especially in cloud environments with numerous virtual machines.
Innovation Solution
A method to display virtual machine disks in a hierarchy, identify parent/child relationships, and consolidate redundant redo logs to reduce chain lengths, thereby improving performance by merging redo logs with only one child node and maintaining snapshot integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If linked cloning is used to create multiple virtual machines sharing a base disk, then disk space is conserved and provisioning speed is improved, but performance degradation occurs when the chain of redo logs increases in length
Solution Approach 1:
The patent merges multiple redo log chains into a single consolidated redo log when they share the same base disk. This consolidation reduces the number of separate redo log files and shortens the effective chain length that the virtual machine I/O stack must traverse, thereby improving performance while maintaining the space-saving benefits of linked cloning
2Adaptability or versatility
If multiple disk chains branch from the same base disk forming a virtual disk tree, then more virtual machines can be created efficiently, but performance degradation becomes noticeable in cloud computing environments with hundreds of linked clones
Solution Approach 1:
The patent merges redo log chains by identifying when multiple virtual disk chains branch from the same base disk and consolidating their redo logs into a single shared redo log. This reduces the complexity of the virtual disk tree structure and shortens the I/O path for all virtual machines sharing that base disk, improving performance while maintaining the ability to create and manage hundreds of virtual machines
Solution Approach 2:
The patent creates a universal redo log that serves multiple virtual disk chains simultaneously. Instead of each virtual disk chain having its own dedicated redo log, a single redo log is shared across multiple chains, reducing redundancy and improving performance while maintaining the linked cloning benefits for multiple virtual machines
Data Source
AI summary
A virtual computer system, including a plurality of virtual machines (VMs) running on one or more host computers, is configured to manage virtual disks in a manner to facilitate making copies of one or more virtual machines (VMs) by means of a method known as linked cloning. Techniques are disclosed for scanning a storage system to determine the set of VMs, identifying virtual disks associated with each VM, examining parent/child relationships among the virtual disks, and displaying the virtual disk hierarchy. Redundant virtual disks may be identified as the set of virtual disks that have exactly one child disk and that are not associated with a snapshot. Provisions for generating a relocate list to support migration of a linked clone may also be utilized.