Hybrid Auto-Scaling Mechanism for Multi-Cloud Resource Coordination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-cloud platforms lack support for global horizontal autoscaling across managed multiple clusters, leading to limited resilience and resource utilization, as well as increased complexity in configuring and managing HPAs across different clouds and clusters.
Innovation Solution
A hybrid horizontal autoscaling mechanism is introduced, comprising a global auto-scaler and one or more local auto-scalers, which collaboratively manage resource scaling across multiple clouds and clusters. The global auto-scaler determines actions based on resource information from multiple clouds and sends modifying requests to local auto-scalers to adjust resource parameters accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If horizontal scaling is performed within a single cluster using HPA, then resource utilization within that cluster is improved, but global resource utilization across multiple clouds remains limited
Solution Approach 1:
The system divides the scaling function into local HPAs for individual clusters and a global HPA for multi-cloud coordination. Each local HPA manages scaling within its own cluster, while the global HPA coordinates across multiple clouds, allowing both local optimization and global adaptability simultaneously.
Solution Approach 2:
The global HPA acts as an intermediary between multiple local HPAs and the multi-cloud environment. It receives metrics from local HPAs, makes global scaling decisions, and distributes scaling actions back to local HPAs, enabling coordinated resource management across clouds while preserving local autonomy.
2Ease of operation
If multiple separate HPAs are configured for each cloud and cluster, then local resource management is achieved, but system complexity increases significantly
Solution Approach 1:
The system merges multiple local HPA functionalities into a single global HPA that coordinates across all clusters and clouds. This consolidation reduces the number of independent scaling controllers needed while maintaining local management capabilities through the global coordinator's oversight.
Solution Approach 2:
The global HPA is designed to perform multiple functions: it monitors metrics from multiple clouds, makes global scaling decisions, coordinates local HPAs, and manages resource distribution across the entire multi-cloud environment. This universal controller simplifies the overall system architecture.
3Speed
If scaling decisions are made independently in each cluster, then local responsiveness is improved, but global resource optimization and resilience are reduced
Solution Approach 1:
The system implements a feedback mechanism where local HPAs continuously report metrics to the global HPA, which then adjusts global scaling decisions based on aggregated information from all clusters. This feedback loop enables both rapid local response and coordinated global optimization, improving overall system resilience.
Solution Approach 2:
The global HPA performs preliminary analysis of metrics from multiple clusters before making scaling decisions. By evaluating the overall system state in advance, it can proactively coordinate scaling actions across clouds, preventing resource imbalances and enhancing global resilience before issues arise.
Data Source
AI summary
Embodiments herein disclose for example a method performed by a scaling arrangement (13) for managing resources in a communication network associated with a multiple cloud infrastructure, wherein the scaling arrangement comprises a global auto-scaler (131) and one or more local auto-scalers (132). The scaling arrangement determines, at the global auto-scaler (131) of the scaling arrangement (13), to perform one or more actions based on at least resource information stored at the global auto-scaler; wherein the resource information is associated with resources structured in multiple clouds comprising one or more local clusters of resources. The one or more actions comprise modifying a parameter related to at least one of the resources of the multiple clouds. The scaling arrangement sends a modifying request to at least one local auto-scaler out of the one or more local auto-scalers (132), requesting to perform one or more operations related to the modified parameter.