Assisted Live Migration of Virtual Machines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional live VM migration processes are inefficient, leading to performance degradation and potential service discontinuity in datacenters, as they consume CPU and network resources and may fail due to incomplete memory state copying during peak utilization.
Innovation Solution
An assisted live migration method where an advisory server monitors workload and uses a load balancer to reduce the load on virtual machines before migration, ensuring the memory state delta is minimal, thereby facilitating efficient and successful VM migration with reduced resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional live VM migration is performed during peak utilization, then workload balancing is achieved, but service continuity and performance are degraded due to incomplete memory state copying
Solution Approach 1:
The system performs preliminary actions by monitoring workload metrics and identifying optimal migration timing before peak utilization occurs. The advisory server analyzes historical and real-time data to schedule migrations during low-utilization periods, preventing the contradiction between maintaining service continuity and achieving successful migration.
Solution Approach 2:
The system implements feedback mechanisms where the advisory server continuously monitors VM performance metrics, resource utilization, and migration outcomes. This feedback loop enables dynamic adjustment of migration scheduling to avoid periods when service continuity would be compromised, thereby resolving the contradiction between reliability and productivity.
2Productivity
If VM migration is performed during high workload periods, then workload balancing is achieved, but resource consumption increases and migration may fail
Solution Approach 1:
The system dynamically adjusts migration scheduling based on real-time workload conditions. The advisory server continuously evaluates resource utilization metrics and adapts migration timing to perform transfers during low-utilization periods, thereby reducing CPU and network resource consumption while maintaining effective workload balancing.
Solution Approach 2:
The system changes operational parameters by shifting migration execution timing from fixed schedules to dynamic, condition-based scheduling. Migration parameters such as source and destination selection, timing, and resource allocation are adjusted based on monitored workload conditions, reducing resource consumption during migration operations.
3Ease of operation
If incremental memory copying is used during live migration, then migration is performed, but service performance degrades due to resource consumption during copying
Solution Approach 1:
The system performs preliminary assessment of memory state delta sizes and predicts migration impact before initiating the migration process. By evaluating memory copying requirements in advance, the system can schedule migrations during periods when service performance degradation would be minimal, thereby maintaining live migration capability while protecting service productivity.
Solution Approach 2:
The advisory server autonomously monitors migration performance and dynamically adjusts migration parameters to minimize service impact. The system self-regulates the migration process by adapting copying rates and timing based on real-time service performance feedback, enabling live migration while maintaining acceptable service levels.
Data Source
AI summary
A method and system for an assisted live migration of virtual machines are provided. The method monitoring, by an advisory server, at least a workload of physical machines in a datacenter; determining if at least one physical machine is overloaded based on the monitored workload; for each of the at least one physical machine determined to be overloaded, selecting at least one virtual machine resides in the respective physical machine, wherein the selection is based at least on a current load of the virtual machine; and initiating a live migration of the selected virtual machine when the current load is lower than a comfort load level.