Cross-Domain Compute Workload Migration via Elastic Management

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Enterprise computing centers face inefficiencies due to excess compute capacity in separate domains like VDI, big data, and cloud, as these domains are designed to meet peak demand, leading to wasted resources and higher costs due to the difference between peak and actual workload demands.

Innovation Solution

A system that employs a monitoring engine and an elastic compute workload management engine to continuously collect and analyze resource utilization data across disparate compute domains, shifting workload from nodes with excess demand to those with available capacity via a high-speed, deterministic network, ensuring real-time optimization of resource utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If separate compute domains are configured with dedicated server nodes to meet peak demand, then reliability is improved, but loss of energy increases due to unused excess capacity

Engineering Contradiction:

Improveservice availabilityVSAvoidwasted compute capacity

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The patent enables server nodes to perform multiple functions by allowing workload migration across domain boundaries. A server node can be dedicated to a specific domain for reliability while simultaneously accepting workloads from other domains when excess capacity exists, making the infrastructure universally applicable to multiple compute domains.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges previously separate compute domains into a unified resource pool. The workload management system combines resources from multiple domains (VDI, big data, grid, cloud) and allows dynamic allocation, effectively combining the excess capacities of individual domains into a shared resource that can be allocated across all domains.

Inventive Principle:

Principle #5Merging (Combining)

2Adaptability or versatility

If domains are built to meet peak demand, then adaptability is improved, but loss of substance increases due to unused resources

Engineering Contradiction:

Improvepeak demand coverageVSAvoidunused compute resources

Core Design Contradiction:

Adaptability or versatilityVSLoss of substance

Solution Approach 1:

The patent introduces dynamic workload allocation that adjusts resource distribution in real-time based on actual demand. The workload management system continuously monitors capacity and demand across domains, dynamically migrating workloads to match current needs rather than static peak-demand allocations, thereby reducing waste while maintaining adaptability.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent implements a mechanism to recover unused compute resources from domains with excess capacity and reallocate them to domains with demand. The system temporarily 'discards' the static domain-bound allocation in favor of a dynamic pooling approach where resources are continuously recovered from underutilized domains and redistributed to meet actual demand.

Inventive Principle:

Principle #34Discarding and recovering

3Device complexity

If dedicated server nodes are used in separate domains, then device complexity is reduced, but productivity decreases due to idle capacity

Engineering Contradiction:

Improvedomain architectureVSAvoidresource utilization

Core Design Contradiction:

Device complexityVSProductivity

Solution Approach 1:

The patent introduces a workload management system as an intermediary layer between dedicated server nodes and compute domains. This intermediary abstracts the complexity of cross-domain resource management, allowing dedicated nodes to remain simple while the intermediary handles the complex logic of monitoring, evaluating, and migrating workloads across domains to optimize utilization.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Ease of operation

If excess capacity is not utilized across domains, then ease of operation is maintained, but loss of time increases due to wasted compute cycles

Engineering Contradiction:

Improvedomain managementVSAvoidwasted compute cycles

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The patent implements self-service automation where the workload management system autonomously monitors capacity, evaluates demand, and migrates workloads without manual intervention. The system automatically detects excess capacity in one domain and reallocates it to domains with demand, eliminating the need for manual resource management while preventing wasted compute cycles.

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS9130844B1Systems and methods for harvesting excess compute capacity across domains

Publication Date: 2015.09.08 CITIGROUP TECHNOLOGY INC

AI summary

Systems and methods for harvesting excess compute capacity across different compute domains involve a monitoring engine that collects resource utilization data continuously in real time for each of a plurality of server nodes in each of a plurality of different compute domains via a network. An elastic compute workload management engine receives the collected resource utilization data from the monitoring engine via the network, extracts excess workload from one or more server nodes with excess workload demand in a first one of the plurality of different compute domains via the network and inserts the extracted excess workload into one or more server nodes with currently available compute capacity in a second one of the plurality of different compute domains via the network.