Adaptive Memory Demand for Workload Placement

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing resource schedulers in high-performance datacenters are ineffective in managing memory contention due to their failure to consider memory reclamation techniques, leading to inefficient workload placement and migration.

Innovation Solution

The implementation of adaptive memory demand mechanisms that account for memory reclamation techniques, such as ballooning and swap memory, to calculate more accurate memory estimates for workload placement and migration, thereby improving resource management.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If resource schedulers use traditional memory demand calculations, then workload placement can be simplified, but memory contention cannot be accurately modeled when memory reclamation techniques are utilized

Engineering Contradiction:

Improvememory demand calculation accuracyVSAvoidresource scheduler complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The memory demand calculation is segmented into multiple components: base memory demand, memory reclamation demand (ballooning, swap, transparent huge pages), and memory overcommitment demand. Each component is calculated separately and then aggregated to provide a comprehensive memory demand estimate that accurately reflects actual memory usage patterns.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary memory reclamation mechanism that acts between the workload and the physical memory. This intermediary layer (through ballooning devices, swap spaces, or transparent huge pages) allows the system to reclaim memory from workloads that are not actively using it, thereby reducing actual memory demand without requiring precise prediction of future memory needs.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If resource schedulers consider memory reclamation techniques in memory demand calculations, then workload placement accuracy improves, but computational overhead increases

Engineering Contradiction:

Improveworkload placement accuracyVSAvoidcomputational energy consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary calculations of memory reclamation potential during idle periods or low-utilization times. By pre-calculating the memory reclamation demand for each workload based on historical usage patterns and current memory state, the system avoids performing these computationally intensive calculations in real-time during workload placement decisions.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent changes the parameters used in memory demand calculation from static allocated memory values to dynamic parameters that reflect actual memory reclamation potential. These parameters include ballooning device usage, swap space utilization, and transparent huge page effectiveness, which provide more accurate indicators of true memory demand.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If memory reclamation techniques are utilized, then memory utilization efficiency improves, but existing resource schedulers become ineffective due to inaccurate memory demand assessment

Engineering Contradiction:

Improvememory utilization efficiencyVSAvoidmemory demand assessment accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system implements feedback mechanisms that continuously monitor actual memory usage versus predicted memory demand. This feedback loop allows the resource scheduler to learn from past performance and adjust memory demand calculations for future workload placement decisions. The feedback also triggers dynamic recalibration of memory reclamation parameters when actual usage patterns diverge from predictions.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11669369B2Cluster resource management using adaptive memory demand

Publication Date: 2023.06.06 VMWARE INC

AI summary

Various examples are disclosed for cluster resource management using adaptive memory demands. In some examples, a local memory estimate is determined for a workload. The local memory estimate is determined using a memory reclamation parameter for the workload executed by a current host of the workload. A destination memory estimate is also determined for the workload. The destination memory estimate is determined using a full memory estimate unreduced by memory reclamation parameters. The workload is executed using a host that is selected in view of an analysis that uses the local memory estimate for the current host and the destination memory estimate for at least one destination host.