VM Migration via UPS Segmentation in Virtual SANs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In virtual Storage Area Networks (SANs), existing systems face challenges in managing Virtual Machines (VMs) effectively when Uninterruptible Power Supplies (UPSs) fail, leading to potential data loss and downtime, as they lack a robust mechanism to differentiate between critical and non-critical UPSs and manage VM migration and shutdown accordingly.
Innovation Solution
A system and method that categorize UPSs into critical and non-critical groups, determine a fault tolerance level, and migrate or shut down VMs based on this classification to ensure continued operation or controlled shutdown, utilizing a computing device to manage hosts and hypervisors, and include user-defined fault tolerance levels for optimal resource management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If VMs are shut down when any UPS fails, then data loss is prevented, but system availability and uptime deteriorate
Solution Approach 1:
The system segments UPS devices into critical and non-critical groups based on their importance to VM operation. Critical UPS failures trigger VM shutdown to prevent data loss, while non-critical UPS failures allow VMs to continue running, thus resolving the contradiction between data protection and system availability.
Solution Approach 2:
Different shutdown policies are applied to different UPS groups. Critical UPS failures result in VM shutdown, while non-critical UPS failures do not. This localized differentiation allows the system to maintain availability for non-critical operations while protecting against data loss for critical operations.
2Reliability
If all UPS failures trigger VM shutdown, then data integrity is maintained, but downtime increases
Solution Approach 1:
The system divides UPS failures into segmented categories (critical and non-critical). Only critical UPS failures trigger VM shutdown and subsequent downtime, while non-critical failures allow continuous operation. This segmentation reduces unnecessary downtime while maintaining data integrity for critical operations.
Solution Approach 2:
The system changes the parameter of UPS importance classification from a single uniform state to a differentiated state (critical vs. non-critical). This parameter change enables selective shutdown behavior that reduces downtime while preserving data integrity where necessary.
3Productivity
If VM migration is implemented for critical UPS failures, then system availability is maintained, but complexity of power management increases
Solution Approach 1:
The system introduces an intermediary classification layer (critical vs. non-critical UPS groups) between the UPS failure event and the VM response action. This intermediary simplifies the decision-making process by providing clear criteria for when to migrate VMs versus when to shut them down, managing complexity through structured classification.
4Reliability
If fault tolerance levels are enforced strictly, then reliability is improved, but flexibility in resource management decreases
Solution Approach 1:
The system implements dynamic resource management where VMs can be migrated between hosts based on real-time UPS status and fault tolerance conditions. When fault tolerance is compromised, the system adaptively relocates VMs to maintain reliability, then dynamically adjusts resource allocation once conditions improve, balancing strict fault tolerance with operational flexibility.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
According to aspects of the present disclosure, a method of operating a virtual storage area network cluster including a plurality of hosts is provided, each host of the plurality of hosts executing at least one Virtual Machine (VM) and being configured to receive power from a respective Uninterruptible Power Supply (UPS) of a plurality of UPSs, the method comprising acts of determining a number of critical UPSs of the plurality of UPSs, each critical UPS of the plurality of UPSs corresponding to a critical host of the plurality of hosts, determining that the number of critical UPSs does not exceed a fault tolerance level, and migrating VMs hosted by the critical hosts to non-critical hosts of the plurality of hosts.