Shared Virtual Storage Manager for HA Cluster Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional clustering solutions, such as Veritas Cluster Server and Microsoft Cluster Server, are not compatible with features like VMotion, Storage VMotion, DRS, snapshots, and VMware HA in virtualization environments like VMware vSphere, limiting high availability clustering in virtual machine setups.
Innovation Solution
A shared virtual storage manager is implemented to facilitate the sharing of virtual storage between multiple virtual machines, enabling failover of applications and VMs without requiring storage media replication, and resolving issues like split brain scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional clustering solutions (Veritas Cluster Server, Microsoft Cluster Server) are used in virtualization environments, then high availability failover is supported, but compatibility with virtualization features (VMotion, Storage VMotion, DRS, snapshots, VMware HA) is lost
Solution Approach 1:
The patent introduces a virtualization environment adapter as an intermediary component that bridges conventional clustering solutions and virtualization platforms. This adapter layer enables the clustering software to interact with virtualized resources through standardized interfaces, maintaining compatibility with VMotion, Storage VMotion, and other virtualization features while preserving failover capabilities. The adapter translates clustering operations into virtualization-aware commands, resolving the incompatibility between traditional clustering and modern virtualized infrastructures.
2Reliability
If storage is shared across all cluster nodes for failover, then application availability is improved, but storage media replication requirements increase complexity
Solution Approach 1:
The patent implements a universal shared storage interface that allows the same storage pool to serve multiple functions simultaneously: primary storage for active applications, standby storage for failover, and replication source for data synchronization. This multi-functional storage architecture eliminates the need for separate replicated storage media, reducing complexity while maintaining high availability. The storage system dynamically allocates resources based on workload requirements, providing both shared access and replication capabilities through a single unified infrastructure.
Data Source
AI summary
Compatibility is provided between a high availability clustering system and a virtualization environment. A hardware level hypervisor runs on each host, and VM(s) run on each hypervisor. A shared virtual storage manager runs on each VM, and facilitates sharing virtual storage between VMs. Sharing virtual storage can be performed in the context of a failed application or VM. A shared virtual storage manager VM detects that the application is being failed over from the source VM to the target VM, detaches the virtual storage from the source VM, and attaches it to the target VM. The shared virtual storage manager can also enable sharing of the virtual storage when a monitored application is failed over in response to the failure of a hypervisor. Virtual storage can additionally act as an arbitrator to resolve an occurrence of split brain, when VMs become disconnected.