Common Volume Storage for Cloud VMs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtualized computing systems face inefficiencies due to the lack of policy-based management at the granularity of a virtual disk in POSIX file systems, leading to complex and less scalable solutions for virtual machine datastores.
Innovation Solution
Implementing a common volume (cVol) that uses a network file system for virtual machine metadata and an object storage system for virtual disks, allowing metadata and data to be scaled independently, and enabling seamless virtualization management workflows with direct, scalable, and performant access to virtual disks under policy-based management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If POSIX file systems are used for virtual machine storage, then compatibility with operating systems is maintained, but policy-based management at virtual disk granularity is lost
Solution Approach 1:
The patent segments the storage system into two independent components: a POSIX file system for metadata (configuration files, logs) and an object storage system for virtual disk data. This segmentation allows each component to be optimized independently - the POSIX system maintains OS compatibility while the object storage system enables policy-based management at virtual disk granularity through separate object-level policies.
Solution Approach 2:
The patent introduces a virtualization layer as an intermediary between the POSIX file system and the object storage system. This intermediary manages the mapping between virtual disk objects and underlying storage resources, enabling policy-based management without requiring changes to the POSIX file system interface, thus maintaining OS compatibility while adding management capabilities.
2Adaptability or versatility
If separate objects are used for each VM configuration file in object datastores, then policy-based management is enabled, but system complexity and scalability are reduced
Solution Approach 1:
The patent merges the management of multiple VM configuration files into a single POSIX file system namespace. Instead of creating separate objects for each configuration file as in traditional object datastores, all VM metadata shares a common file system structure, reducing the number of separate objects while maintaining policy-based management capabilities through the virtualization layer.
Solution Approach 2:
The POSIX file system serves multiple functions simultaneously: it stores VM configuration files, manages metadata, and provides the namespace structure for the virtualization layer. This multi-functionality eliminates the need for separate object structures for each VM configuration, simplifying the overall datastore architecture while preserving policy-based management through the virtualization layer's object abstraction.
3Ease of operation
If shared storage is used for virtual machine data, then data accessibility is improved, but cluster scalability is limited
Solution Approach 1:
The patent segments storage access into two independent paths: metadata access through the POSIX file system and virtual disk data access through the object storage system. This segmentation allows the object storage system to scale independently to support larger clusters while the POSIX file system handles metadata operations, maintaining data accessibility without being constrained by shared storage scalability limits.
Solution Approach 2:
The patent replaces traditional mechanical shared storage systems with a software-defined storage architecture using object storage. This substitution enables scalability through software rather than hardware constraints, allowing the cluster to expand beyond the limits of physical shared storage while maintaining accessibility through the virtualization layer's abstraction.
Data Source
AI summary
An example method of providing a common volume (cVol) datastore for virtual machines (VMs) managed by a hypervisor in a cloud computing system includes: mounting, by the hypervisor in cooperation with a network file system server, a network file system share of a common volume (cVol), the network file system share storing metadata for the VMs; creating a file system container backed by the network file system share; routing file operations targeting the metadata to the file system container; attaching cloud volumes as devices on a host of the hypervisor, the cloud volumes referenced by descriptors in the metadata; and routing file operations targeting virtual disks of the VMs to the devices.