Acceldata 26.4.0 delivers a significantly enhanced dbt Cloud Connector — addressing core reliability issues, adding enriched metadata context, and delivering a unified pipeline view that accurately reflects how dbt Cloud jobs execute.
dbt (Data Build Tool) has become the de facto standard for SQL-based data transformations inside modern data warehouses like Snowflake, BigQuery, and Databricks. Its cloud SaaS offering, dbt Cloud, is widely adopted in enterprise environments — including Workday, whose requirements directly shaped this GA milestone. Unlike dbt Core, dbt Cloud does not natively support Open Lineage, making a purpose-built integration essential for teams that need observability across their transformation pipelines.
Why This Matters
The original dbt Cloud connector relied on Open Lineage events emitted per command step within a job. This approach introduced several critical gaps:
- A single dbt Cloud job execution containing multiple commands would generate multiple separate pipeline runs in Acceldata — creating confusion and operational noise.
- There was no unified lineage view of what a complete job run actually did end-to-end.
- Metadata context for individual model and test executions — including the compiled SQL queries behind them — was missing.
- There was no data source association, meaning users could not filter or link dbt pipelines to registered data sources in Acceldata.
Key insight: Fragmented pipeline runs made it nearly impossible to answer a simple question: what did this dbt job do, and did it succeed?
What’s New in the GA Release
Holistic Pipeline Representation
Each dbt Cloud job is now represented as a single unified pipeline in Acceldata. A job run maps directly to a pipeline run — eliminating the step-level fragmentation that caused multiple runs to appear for a single execution.
In-house Lineage Stitching from dbt Artifacts
Rather than relying on Open Lineage events, the new implementation parses dbt's native artifacts directly — manifest.json, run_results.json, and sources.json — to reconstruct accurate, end-to-end lineage. This approach gives Acceldata full control over lineage quality without depending on external event emission.
Execution-based DAG Processing
Only the nodes relevant to a specific job run are processed. If a project contains many models but a given job only touches a subset, the pipeline graph reflects exactly what executed — not the full project definition. This improves performance and keeps the lineage view focused and actionable.
Full dbt Resource Type Support
The connector now supports all dbt resource types with proper Acceldata representation:
- Models — rendered as a Job Node that materializes into an Asset Node (view or table)
- Snapshots — capture slowly changing dimensions; appear as Asset Node → Job Node → Asset Node
- Seeds — load CSV data into warehouse tables; follow the same Asset → Job → Asset pattern
- Sources — represent raw tables used for freshness measurement
- Tests — appear as Job Nodes that evaluate data quality without creating assets
Test Failure Visibility
When a dbt test fails, Acceldata surfaces the failure directly in the pipeline view — including the number of affected rows and the compiled SQL query responsible for the failure. This provides teams with the context needed for fast root cause analysis, without leaving the observability platform.
Circuit Breaker Behavior
dbt Cloud natively skips downstream steps when an upstream node fails. Acceldata now reflects this behavior accurately — showing skipped nodes in their correct state within the pipeline run view. This gives operators a clear picture of failure propagation and helps distinguish between nodes that actively failed versus those that were bypassed due to upstream issues.
Data Source Association and Filtering
dbt Cloud can now be registered as a data source in Acceldata. Once onboarded, pipelines associated with that integration are filterable by data source and asset — bringing dbt Cloud into the same operational workflow as all other monitored sources.
Selective Project Onboarding
During setup, users can choose which dbt Cloud projects to monitor. Rather than ingesting an entire account, teams can scope Acceldata monitoring to the specific projects that matter most to their workflows.
dbt Concepts, Mapped to Acceldata
For teams new to the integration, here is how dbt Cloud terminology maps to Acceldata concepts:
- dbt Cloud Job → Acceldata Pipeline
- dbt Cloud Job Run → Acceldata Pipeline Run
- dbt Model → Job Node → Asset Node
- dbt Snapshot / Seed → Asset Node → Job Node → Asset Node
- dbt Test → Job Node (no output asset)
- dbt Source → Asset Node → Job Node
Who This Is For
The dbt Cloud GA connector is built for data engineering and analytics engineering teams that:
- Run transformation workloads on dbt Cloud against Snowflake, BigQuery, Databricks, or other supported warehouses
- Need end-to-end pipeline visibility across their ELT stack — from raw ingestion through transformation to consumption
- Want test failure alerts and lineage context without switching between tools
- Operate in enterprise environments where reliability and auditability of transformation pipelines are critical
Available Now
The dbt Cloud Connector GA is available as part of the Acceldata 26.4.0 release. Existing preview configurations carry forward without requiring migration. Reach out to your Acceldata Customer Success contact or visit the documentation portal to begin onboarding dbt Cloud projects.
.webp)
.webp)
