ALL-DATA0011 · unowned_high_impact_asset
Count assets nobody owns that many downstream assets depend on.
This is a deterministic rule for all languages. Read its implementation.
Definition
Section titled “Definition”Walk the lineage graph outward from every asset, up to maximum_depth hops, count the distinct
assets it reaches, and report an asset the catalog gives no owner when that count reaches
minimum_downstream. Ownership is what turns a schema question into a conversation, and the
assets that most need one are the ones a change propagates furthest from. An unowned leaf costs
one team an afternoon while an unowned root stops a reporting stack.
Impact is counted from the graph rather than guessed from a name. A table named curated may
feed nothing while one named scratch feeds the whole warehouse, so the arrows say which is
which and the naming convention says nothing.
Evidence
Section titled “Evidence”Each finding names the unowned asset and how many distinct assets sit downstream of it within the walked depth. The value is the number of unowned assets whose downstream reach meets the threshold.
Exceptions
Section titled “Exceptions”An asset naming any owner is never reported however far its lineage reaches, since this asks
who answers rather than whether the answer is the right one. An asset whose reach falls under
minimum_downstream stays quiet, which is what keeps a catalog of small unowned staging tables
from drowning the roots that matter. An asset reachable only through a lineage edge the
snapshot never recorded counts as unreached, so ALL-DATA0010 is the rule that says whether
this graph is complete enough to trust. A cycle contributes each asset once because the reach
is a distinct set, and an asset never reaches itself.
Examples
Section titled “Examples”An unowned raw.orders feeding staging.orders, which feeds mart.revenue and
mart.invoices, reaches three assets and returns 1. The same asset with one named owner
returns 0, and so does an unowned asset feeding only two others.
References
Section titled “References”- Cites “OpenLineage specification”, object model
- Cites “DAMA-DMBOK”, data stewardship and accountability
- Cites “DataHub documentation”, ownership and lineage metadata