Workflow-Aware NUMA Scheduling for Virtual Machine Lifecycle Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of workload scheduling in NUMA systems increases due to multiple processor sockets and varying CPU and memory loads, leading to inefficiencies in virtual machine management, particularly when optimizations do not benefit performance and incur additional costs.
Innovation Solution
Implementing a workflow-aware NUMA optimization method that identifies specific trigger conditions to adjust optimization operations, such as skipping memory migrations, calculations, or optimizations entirely for virtual machines or containers that are about to be terminated or migrated, thereby reducing unnecessary performance impacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If NUMA optimization operations are performed for all virtual machines, then memory access performance is improved, but system complexity and overhead increase
Solution Approach 1:
The patent changes the parameter of optimization application by introducing workflow-awareness as a conditional parameter. The NUMA optimizer monitors workflow states and adjusts its behavior based on whether a virtual machine is in an active workflow or a transitional state (migrating, shutting down, paused). This selective application of optimizations based on workflow parameters resolves the contradiction by applying complex optimizations only when beneficial.
Solution Approach 2:
The patent implements dynamic adjustment of NUMA optimization behavior based on real-time workflow states. The system transitions from static optimization (always applying NUMA optimizations) to dynamic optimization (adjusting optimization intensity based on workflow awareness). This allows the system to adapt optimization levels to current operational conditions, reducing unnecessary complexity while maintaining performance benefits.
2Speed
If NUMA optimizations are applied to virtual machines about to be migrated or shut down, then memory access performance is improved, but unnecessary performance penalties are incurred
Solution Approach 1:
The patent applies preliminary action by detecting workflow state changes (migration, shutdown, pause) before NUMA optimization operations are executed. The workflow-aware NUMA optimizer identifies transitional states in advance and prevents the initiation of optimization operations during these periods. This preliminary detection and prevention mechanism avoids the performance penalties that would result from unnecessary optimizations during transitional states.
Solution Approach 2:
The patent implements skipping by intentionally bypassing NUMA optimization operations during specific workflow states. When the system detects that a virtual machine is in a transitional state (migrating, shutting down, or paused), it skips the optimization operations entirely. This selective skipping prevents the incurrence of unnecessary performance penalties while maintaining optimizations during stable operational states.
3Productivity
If comprehensive NUMA optimization operations are performed, then resource allocation efficiency is improved, but calculation overhead increases
Solution Approach 1:
The patent changes the operational parameters of the NUMA optimizer by introducing workflow-awareness as a controlling parameter. Instead of performing comprehensive optimization operations on all virtual machines regardless of state, the system adjusts its operation parameters based on workflow detection. This parameter-based control reduces calculation overhead by focusing optimization efforts only on virtual machines in appropriate states.
Solution Approach 2:
The patent applies partial action by performing NUMA optimization operations only on a subset of virtual machines that are in active workflow states, rather than applying optimizations comprehensively to all virtual machines. This selective partial application of optimization operations maintains resource allocation efficiency for critical workloads while reducing the overall calculation overhead by excluding virtual machines in transitional or inactive states.
Data Source
AI summary
Disclosed herein are techniques for performing workflow-aware NUMA (non-uniform memory access) optimizations. “Execution contexts” such as virtual machines or Docker containers are managed within a computer system that includes multiple NUMA nodes, each including a processor socket and memory local to that socket. NUMA optimization techniques are performed whereby execution contexts and/or data for such contexts are migrated between NUMA nodes. Migration may occur for various reasons, including for load balancing, to achieve fairness, or to increase the amount of data that is stored in memory local to a particular execution context. When it is known that an execution context will soon stop executing in a particular computer system, optimizations may provide little or no benefit. In such situations, the computer system performs modified NUMA optimizations, in which certain optimizations that would normally be performed are instead not performed.