High-Availability Power Management for Virtual Desktop VMs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face challenges in ensuring high availability of virtual machines (VMs) during peak usage periods due to failures in power and capacity management (PCM) services, leading to potential outages and increased costs from requiring local management server setups.
Innovation Solution
A lightweight, high-availability power management service is deployed within a customer's data center or cloud system as a stateless container, which receives credentials and VM lists from a PCM service and autonomously powers on VMs if the PCM service fails to manage them according to specified criteria, ensuring continuous operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a local copy of the management server is hosted within the customer's data center to ensure high availability during PCM service failures, then VM availability is improved, but maintenance effort and resource usage costs increase
Solution Approach 1:
The patent extracts only the essential power management functionality from the full management server, creating a lightweight HA service that performs specifically health checks and VM power operations. This extraction maintains reliability while reducing complexity and resource requirements compared to deploying a complete management server copy.
Solution Approach 2:
The HA service is designed as a lightweight, stateless container that can be rapidly deployed and replaced. It uses minimal resources and can be quickly provisioned as needed for high availability, rather than maintaining a permanent, resource-intensive local management server installation.
2Loss of energy
If cloud system auto scaling is used to power on and off computing resources based on load, then usage costs are reduced, but VM availability during peak periods deteriorates when PCM service fails
Solution Approach 1:
The HA service performs preliminary health checks on the PCM service to detect potential failures before they impact VM availability. By proactively monitoring PCM service status and pre-positioning itself with necessary credentials and VM information, the system can rapidly respond to failures without waiting for actual outages, thus maintaining both cost efficiency and reliability.
Solution Approach 2:
The HA service acts as an intermediary between the cloud system's auto scaling infrastructure and the PCM service. It receives credentials and VM lists from PCM, monitors PCM health, and can directly interface with the hypervisor to power on VMs when needed, bridging the gap between cost-saving auto scaling and reliable VM availability.
3Productivity
If the PCM service selectively powers on and off VMs according to criteria, then resource optimization is improved, but system complexity increases requiring additional health check mechanisms
Solution Approach 1:
The system segments the power management functionality into distinct components: the PCM service that handles selective VM power management based on criteria, and the HA service that handles health monitoring and failover. This segmentation allows resource optimization through PCM while isolating the complexity of health checks and failover logic to a dedicated HA component, reducing overall system complexity.
Data Source
AI summary
A method can include receiving, by a high-availability (HA) service running in a data center having a plurality of virtual machines (VMs) and a hypervisor, credentials for the hypervisor and a list of VMs, the hypervisor credentials and the list of VMs received from a power and capacity management (PCM) service, the PCM service configured to selectively power VMs from the list of VMs on and off; sending, from the HA service to the PCM service, a health check request to determine if the PCM service is able of selectively power the VMs from the list of VMs on and off; and in response to determining the PCM service is unable to selectively power the VMs from the list of VMs on and off, powering on, by the HA service, one or more of the VMs from the list of VMs using the hypervisor credentials received from the PCM service.