Observability

Observability in Spice enables task tracking and performance monitoring through a built-in distributed tracing system that can export to Zipkin or be viewed via the runtime.task_history SQL table.

Spice records detailed information about runtime operations through trace IDs, timings, and labels - from SQL queries to AI completions. This task history system helps operators monitor performance, debug issues, and understand system behavior across individual requests and overall patterns.

Use-Cases

Debugging and Troubleshooting

• Trace AI chat completion steps and tool interactions to identify why a request isn't responding as expected
• Investigate failed queries and other task errors

Performance Analysis

• Track SQL query/tool use execution times
• Identify slow-running tasks

Usage Analytics

• Track usage patterns by protocol and dataset
• Understand how AI models are using tools to retrieve data from the datasets available to them

Portal Interface

The Spice platform provides a built-in UI for visualizing the observability traces that Spice OSS generates.