A cloud-native framework for seismic waveform data quality assessment: Performance evaluation under equivalent resource constraints.

Abstract

The rapid growth of seismic waveform data requires efficient and elastic processing, but virtual-machine-based architectures often tie workload execution to coarse resource units and rigid scaling policies. To address these limitations, we propose a Kubernetes- and Docker-based cloud-native framework for real-time seismic waveform data quality assessment. The framework organizes the processing logic into three tiers: infrastructure and orchestration, elastic scaling policy, and data quality assessment services. It further introduces an Adaptive Elastic Scheduling Algorithm (AESA) that combines CPU and memory high-watermark triggers with task-lifecycle-aware contraction for containerized computation of key quality indicators, including gap count, gap duration, and percent of availability. Under equivalent hardware resource constraints, the cloud-native implementation reduced mean response latency by 47.27% at 20 requests per second and 47.26% at 50 requests per second compared with the VM-based control group, corresponding to an approximately 1.90-fold speedup in both load regimes. These results indicate that, for the tested gap-count, gap-duration, and availability workloads, cloud-native orchestration can improve latency without changing the quality-metric code. The evaluation is limited to the specified workload, resource budget, and two traffic levels.