Why Observability Needs a Context Graph
Snowflake introduces an Observability Context Graph in Observe to transform raw telemetry into queryable, connected data sets for faster incident investigations across services and dependencies.
Modern observability systems produce vast amounts of telemetry, yet this data often fails to reveal how system components interact during incidents. The new Observability Context Graph in Observe by Snowflake addresses this by converting raw telemetry into structured, queryable data sets that map relationships across services, infrastructure, and dependencies. This allows teams to trace not just what broke but why, replacing manual event correlation with a connected view of their environment. The graph sits atop the Telemetry Data Lakehouse, centralizing all observability data for unified access and analysis.
The context graph organizes telemetry into nodes (data sets) and edges (links and lineage), with correlation tags providing shared metadata to navigate across datasets. These connections enable investigations to span multiple data sets simultaneously rather than in isolation. The AI SRE leverages these relationships to identify relevant data, helping teams diagnose root causes of incidents such as service outages or deployment impacts without piecing together disparate events manually.
Users can explore the graph through Observe’s interface, including schema inspection, correlation tags, and relationship views like Links, Focus, and Lineage. These tools reveal data set connections and origins, while APIs allow programmatic traversal of the graph for automated or agent-assisted investigations. The APIs return precise, token-efficient results by resolving search terms to correlation tags, then locating associated data sets and metrics, even in large-scale environments.
To utilize the Observability Context Graph, infrastructure data must first be routed to the Telemetry Data Lakehouse via an integration, Observe Agent, or OpenTelemetry collector. Once connected, the graph encodes observability expertise, enabling targeted investigations through APIs that map service relationships and return contextualized metrics. This reduces manual effort by providing direct answers and next-step guidance based on the underlying data.