Observability for data platforms: From pipelines to platform SLOs

Vineel Arekapudi discusses how enterprise data platforms require observability beyond basic logs and metrics. The session explores how Spark jobs, orchestration pipelines, and Kubernetes telemetry at Wells Fargo were unified into platform-level SLOs that improve reliability, cost visibility, and developer trust.

Key takeaways:
• Instrumenting Spark and pipelines for platform observability
• Mapping technical signals to platform SLOs
• Improving reliability and adoption with actionable telemetry
• Common monitoring gaps in large-scale data platforms