Operations | Monitoring | ITSM | DevOps | Cloud

AI applications can behave unpredictably, potentially leading to errors such as hallucinations or data leaks, even when classic monitoring indicates a successful response. To effectively monitor AI systems, four key areas should be focused on. Implementing these pillars can enhance trust in AI deployments, help manage costs, and identify safety issues before they impact users.

A semantic-convention rename (http.method → http.request.method) can silently break your RED metrics — no errors, just gaps in dashboards and alerts. The OpenTelemetry Collector's schema processor fixes it: put it first in your pipeline and it normalizes attribute names no matter what each service emits. Migration mode writes BOTH the old and new names, so you get zero-downtime upgrades while queries keep working.

This walk-through shows how Splunk's Cloud can discover schema and partition keys for Microsoft Azure Blob Storage datasets and create searchable Splunk managed tables. Once the data is mapped, analysts can use Splunk Federated Search to query Azure Blob data where it lives, bringing cloud-resident logs into security, observability, and operational work-flows without re-ingesting the data.

This walk-through shows how Splunk's Cloud can discover schema and partition keys for Microsoft Azure Blob Storage datasets and create searchable Splunk managed tables. Once the data is mapped, analysts can use Splunk Federated Search to query Azure Blob data where it lives, bringing cloud-resident logs into security, observability, and operational work-flows without re-ingesting the data.

This walk-through shows how Splunk Cloud can search AWS S3 data through an Apache Iceberg REST catalog backed by Nessie. Learn how Iceberg table metadata, S3 storage, and Splunk Federated Search work together so analysts can query historical security data where it lives without reingesting it into Splunk.

Government payment fraud is a fast-growing risk for public sector organisations in Australia and globally. From welfare and healthcare payments to business grants and disaster relief, increasingly sophisticated organised criminal networks and other actors exploit complex, high-volume government programs to unlawfully access public funds. The impact is significant—billions lost, program integrity undermined, and essential resources diverted.

This walk-through shows how Splunk's crawler, available through the Data Management app, can discover schema and partition keys for S3 backed datasets and create Splunk managed catalog tables. Once the data is mapped, analysts can search AWS S3 data through Splunk and bring it into broader security, observability, and operational workflows.

Are storage costs and data silos slowing down your investigations? In this video, we dive into the Unified Dataset Experience to show you how to search data where it lives. Learn how to use the Splunk Ingest Processor to route high volume logs directly to AWS S3 while maintaining instant visibility via Federated Search. No more re-hydrating data, just fast cost-effective insights.

In this video, I walk through the three main pieces of the new Kubernetes monitoring experience in Splunk Observability Cloud: the Kubernetes overview page for monitoring the status and top issues across your environment, the Kubernetes Entities page for troubleshooting individual instances with correlated metrics, logs, events, and configuration, and the Workload Optimization view for getting actionable recommendations on your CPU and memory resource allocation.