AI Load Balancing for Sustainable Hybrid Cloud Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods do not optimally route traffic at the network layer to optimize for sustainability, particularly in hybrid cloud environments, neglecting the environmental component of Environmental, Social, and Governance (ESG) considerations.
Innovation Solution
Implement a sustainability aware load balancer that collects power consumption metrics, builds a machine learning regression model, and selects target servers based on minimal power consumption while meeting service level agreements to route transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If traditional load balancing methods are used, then traffic routing is simple and fast, but sustainability metrics (power consumption, carbon emissions) are not optimized
Solution Approach 1:
The system performs preliminary actions by collecting power consumption metrics from server endpoints before making routing decisions. The machine learning regression model is trained in advance on historical power consumption data, enabling the load balancer to predict and select energy-efficient routing paths without adding real-time computational complexity to the decision-making process.
Solution Approach 2:
The patent introduces an intermediary machine learning regression model that acts as a mediator between traditional load balancing and sustainability optimization. This model translates complex power consumption metrics into routing decisions, allowing the system to optimize energy usage without requiring direct complex interactions between multiple sustainability parameters and routing logic.
2Reliability
If power consumption metrics are collected and ML models are built for every server, then sustainability optimization is achieved, but system complexity and computational overhead increase
Solution Approach 1:
The machine learning regression model serves multiple functions: it predicts power consumption for routing decisions, validates service level agreement compliance, and adapts to changing server characteristics. This multi-functionality reduces the need for separate systems for each task, thereby managing complexity while achieving reliability goals.
Solution Approach 2:
The system implements feedback mechanisms where power consumption metrics collected from server endpoints are continuously fed into the machine learning model. This feedback loop allows the model to learn from actual power consumption patterns and improve routing decisions over time, ensuring service level agreements are met while optimizing sustainability metrics.
Data Source
AI summary
A method, system, and computer program product for load balancing by a sustainability aware load balancer are provided. The load balancer collects power consumption metrics for each server endpoint in a topology. The power consumption is based on a workload and characteristics of a plurality of transactions on the server. A machine learning regression model based is built on the collected power consumption metrics. A target server is selected for the transaction, based on the machine learning regression model identifying the server consuming the least amount of power while meeting service level agreements, and the transaction is routed to the target server.