databricks Platform · Updated Jun 11, 2026

Endpoint Telemetry for Custom Model Serving

Now generally available for enhanced observability.

What it does

This feature allows users to configure endpoint telemetry for custom model serving endpoints, enabling the persistence of OpenTelemetry logs, traces, and metrics to Unity Catalog tables. It supports root cause analysis, monitoring of endpoint health, and compliance with SQL queries. Users can instrument their model code to capture telemetry data seamlessly.

Why it matters

By providing detailed telemetry data, organizations can enhance their model performance monitoring, improve troubleshooting capabilities, and ensure compliance with data governance standards. This leads to better decision-making, reduced downtime, and increased trust in model outputs.

Key capabilities

Telemetry Data Persistence

Persist OpenTelemetry logs, traces, and metrics to Unity Catalog tables.

Root Cause Analysis

Utilize telemetry data for effective root cause analysis of model performance issues.

Health Monitoring

Monitor the health of model serving endpoints through detailed telemetry insights.

Compliance Support

Meet compliance requirements with standard SQL queries on telemetry data.

Prerequisites

Workspace must be enabled for Unity Catalog.

Permissions: USE CATALOG, USE SCHEMA, CREATE TABLE, and MODIFY on the destination Unity Catalog.

An existing custom model serving endpoint or permissions to create one.

Workspace must be in a supported region.

Recommended use cases

Monitoring model performance over time.Identifying and troubleshooting errors in real-time.Ensuring compliance with data governance policies.Analyzing usage patterns for model endpoints.

Limitations · remember

!Schema evolution on the target table is not supported.

!Only managed Delta tables are supported.

!The table location must be in the same region as your workspace.

!Only single availability zone durability is supported.

!Telemetry latency degrades beyond 2500 QPS.

Architecture

Custom Model Serving Endpoint → Telemetry Configuration → OpenTelemetry Logs → Unity Catalog Tables

Get started

Instrument Model Code

Add logging and optional OpenTelemetry instrumentation to your model code.

Prepare Unity Catalog

Ensure a catalog and schema are ready to receive telemetry data.

Enable Telemetry

Configure telemetry settings when creating or updating an endpoint.

Verify Telemetry Data

Check the Unity Catalog tables to ensure data is flowing correctly.

Query Telemetry Data

Run SQL queries to analyze the telemetry data stored in Unity Catalog.

Key takeaway

Endpoint telemetry enhances observability for custom model serving, enabling better monitoring and compliance.

2026-06-11-endpoint-telemetry-for-custom-model-serving-endpoints-is-generally-available Generated by the Databricks Updates Agent · browse all updates