Study Databricks DA-ASSOC Lineage and Discovery: key concepts, common traps, and exam decision cues.
Analysts lose time and trust when they use the wrong asset or cannot explain where a metric came from. This part of the exam checks whether you can use metadata, tags, and lineage to work like a governed analyst rather than a guess-and-check user.
| Tool or signal | Why it matters |
|---|---|
| tags and metadata | help analysts find and classify relevant assets |
| certification signal | indicates higher trust for reuse |
| lineage | validates upstream sources and downstream impact |
| ownership | tells you who controls or curates the object |
| Signal in the stem | Best first instinct |
|---|---|
| “which dataset should the analyst use?” | prefer the governed asset with the clearest trust indicators |
| “why do users question the metric?” | inspect lineage, ownership, and source semantics |
| “tag the asset” | think discoverability and classification, not row-level SQL logic |
| Trap | Better rule |
|---|---|
| using lineage only after results look wrong | lineage is also a discovery and validation tool before analysis |
| assuming tags replace documentation or governance | tags help discovery, but trust still depends on ownership, lineage, and semantics |
| choosing the newest table without checking trust | freshness is not the same as trustworthiness |
Discovery questions usually hinge on whether you are validating trust or improving findability. If the goal is “where did this come from,” think lineage. If the goal is “how do people find and classify this,” think descriptions and tags. The weak answer usually confuses monitoring or compute features with governance metadata.