Unified SLO Driven Observability across AWS and GCP using CloudWatch and Stack driver and ELK and Splunk with Error Budgets

Abstract The rapid adoption of multi-cloud architecture has intensified challenges in achieving consistent reliability governance across heterogeneous cloud providers and observability platforms. Although AWS and GCP offer mature monitoring tools, and enterprises integrate systems such as ELK and Splunk, reliability management remains fragmented across provider-specific metrics and alerting semantics. This limits cross-cloud SLO enforcement and obscures error-budget consumption. This study proposes and evaluates a unified SLO-driven observability framework that integrates CloudWatch, Cloud Monitoring, ELK, and Splunk under a centralized SLO abstraction and error-budget governance layer. A provider-independent SLO meta-model normalizes telemetry semantics, while cross-cloud burn-rate computation is embedded within CI/CD control loops. Experimental evaluation in a controlled AWS–GCP environment shows significant improvements over siloed monitoring, including a 60% reduction in mean time to detection, a 40% reduction in mean time to recovery, improved SLO stability, and an 18% reduction in operational cost. The findings indicate that multi-cloud reliability fragmentation is primarily a governance and semantic integration issue. The proposed framework provides a scalable foundation for unified and economically optimized multi-cloud reliability management.