Heterogeneous VM Deployment With Greedy Zone Balancing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional VM deployment systems struggle to provide optimized recommendations for heterogeneous VMs across multiple regions and availability zones, often leading to network resource strain and inefficient deployment due to biased initial scoring, which can result in inaccessible VMs or costly migrations.

Innovation Solution

A computerized method using a greedy algorithm with a recommendation timer to quickly generate balanced VM distributions by evaluating high-scoring VMs first, adjusting weights for load balancing, and deploying heterogeneous VMs across multiple regions and availability zones within a constrained time frame.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional VM deployment systems use biased initial scoring, then deployment speed is improved, but deployment optimization and load balancing deteriorate

Engineering Contradiction:
Improvedeployment speedVSAvoiddeployment optimization
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The system dynamically adjusts VM deployment strategies by transitioning from static biased initial scoring to dynamic greedy algorithms that adaptively evaluate candidate VMs and update location usage counters in real-time, optimizing deployment based on current system state

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The invention changes the scoring parameters from fixed biased values to dynamic scores generated by greedy algorithms that consider multiple factors including location usage thresholds, load balancing requirements, and optimization goals, thereby improving both speed and optimization quality

Inventive Principle:
Principle #35Parameter changes

2Ease of operation

If VMs are concentrated in single regions, then deployment simplicity is improved, but network resource strain and availability deteriorate

Engineering Contradiction:
Improvedeployment simplicityVSAvoidVM availability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The system applies local quality by evaluating and optimizing VM placement in specific locations individually, using location usage counters and greedy algorithms to distribute VMs across multiple regions and availability zones based on local conditions, preventing concentration in single regions

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The invention segments the deployment problem into location-specific sub-problems, evaluating candidate VMs for different locations separately and using location usage thresholds to control distribution across multiple regions, thereby improving availability while maintaining operational simplicity

Inventive Principle:
Principle #1Segmentation

3Manufacturing precision

If comprehensive VM evaluation is performed, then deployment optimization is improved, but processing time and resource requirements deteriorate

Engineering Contradiction:
Improvedeployment optimizationVSAvoidprocessing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs partial evaluation by using greedy algorithms to evaluate only the most promising candidate VMs first, updating location usage counters incrementally, and stopping when optimization criteria are met or time expires, rather than exhaustively evaluating all possible deployments

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The invention implements time constraints and recommendation timers to rush through the evaluation process, skipping less promising candidate VMs and locations, and providing optimized recommendations within constrained time frames rather than performing complete comprehensive evaluation

Inventive Principle:
Principle #21Skipping (Rushing through)

Data Source

PatentEP4625170A1Systems and methods for optimizing VM deployments
Publication Date: 2025.10.01 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP4625170A1 patent drawingFigure 1
  • EP4625170A1 patent drawingFigure 2
  • EP4625170A1 patent drawingFigure 3

AI summary

A computerized method optimizes VMs (VMs) deployment. Optimization goals are received for a heterogeneous deployment of a group of VMs. A recommendation timer is set and candidate VMs are identified based on the optimization goals. When the recommendation timer has not expired: a score for each of the identified candidate VMs is determined, a candidate VM with a highest score amongst the determined scores is assigned to a first group recommendation of group recommendations, and a location usage counter for the location of the candidate VM is updated. If the location usage counter meets a threshold, candidate VMs having similar location are removed from further calculations, otherwise, scores for the candidate group of VMs is recomputed. Based on a total score for each of the group recommendations, a group recommendation is selected to deploy the group recommendation comprising the group of heterogeneous VMs.