Dynamic Compute Adaptation Using Observability Metrics in Hybrid Cloud Architectures
Year: 2025
Type: article
Abstract: The article explores an architecture for dynamically adapting compute resources in hybrid cloud environments based on real-time observability metrics. It presents an approach that leverages both traditional server-based infrastructure and event-driven serverless computing to efficiently handle varying workloads while optimizing cost and performance. The article details a comprehensive observability framework that continuously monitors system metrics, application performance, and business outcomes to make intelligent resource allocation decisions. Implementation techniques, including queue-based load management, warm-start optimization for serverless functions, and multi-layered fallback mechanisms, ensure seamless operation during scaling events. The architecture proves particularly valuable in domains with variable workload patterns such as retail personalization, predictive maintenance, and digital healthcare, where organizations must balance cost efficiency with performance reliability. Experimental results demonstrate that this hybrid model significantly reduces baseline compute costs while maintaining system availability during traffic spikes. The article offers practical guidance for engineers building elastic AI systems in cost-sensitive, latency-bound domains.
Author: Akash Goel
Institution: Westcliff University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6912083490
FWCI:
Cites:
Cited by:
Related to: