Workload-Aware vCPU Placement to Reduce Cache Contention

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing virtualized environments often default to distributing virtual CPUs (vCPUs) across different modules, which may not optimize performance for various workloads, leading to inefficiencies in cache utilization.

Innovation Solution

Implementing a configuration policy that maps workload types to specific shared cache placement modes, including strict and relaxed shared modes, to assign vCPUs to hardware processing cores based on workload requirements, allowing for flexible and efficient cache resource utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If vCPUs are distributed across different modules by default, then device complexity is reduced and ease of operation is improved, but cache utilization efficiency deteriorates and productivity decreases

Engineering Contradiction:

Improvecomputational efficiencyVSAvoidcache placement management

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system performs preliminary classification of workloads into cache-friendly and cache-unfriendly types before vCPU placement. This advance categorization enables the placement policy to pre-determine optimal cache placement strategies, avoiding runtime performance degradation and improving overall computational efficiency without adding complex runtime management overhead

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention changes the placement parameter from uniform distribution to workload-type-dependent placement. By modifying the placement policy parameter based on workload classification results, the system optimizes cache utilization for different workload types, achieving up to 9% performance improvement for server-side Java and 4% for AI inference tasks

Inventive Principle:

Principle #35Parameter changes

2Productivity

If vCPUs are uniformly distributed across modules, then ease of operation is improved, but cache contention increases and performance deteriorates

Engineering Contradiction:

ImproveperformanceVSAvoidvCPU placement management

Core Design Contradiction:

ProductivityVSEase of operation

Solution Approach 1:

The system applies different placement strategies to different workload types locally. Cache-friendly workloads are placed to utilize shared caches effectively, while cache-unfriendly workloads are placed to minimize cache contention. This localized optimization approach improves overall system performance without requiring complex global management, as each workload type receives tailored placement treatment

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

Workload classification is performed in advance before vCPU placement, enabling the system to pre-determine the appropriate placement strategy for each workload type. This preliminary action eliminates the need for complex runtime adjustments and simplifies placement management while achieving significant performance improvements

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250298654A1Non-transitory machine-readable storage medium, method and apparatus for placement of virtual processors

Publication Date: 2025.09.25 INTEL CORP

AI summary

Techniques are disclosed for improving virtual machine performance through vCPU placement. A configuration policy maps workload types to shared cache placement modes, including strict and relaxed modes. A workload type associated with a virtual machine (VM) is identified, and a corresponding cache placement mode is selected based on the policy. Virtual CPUs (vCPUs) of the VM are then assigned to hardware processing cores according to the selected mode, enabling optimized use of shared cache resources based on workload characteristics and system topology.