.svg){kind=icon}
Learn
A data catalog is an organized inventory of an organization's data assets that records each asset's metadata, location, meaning, lineage, and access rules so people and systems can discover, understand, trust, and reuse data without first owning it.
Last updated: October 28, 2026
A sustainment shortfall is usually visible in the data days before it is visible on the flight line. The fuel consumption rates, the parts backorders, and the maintenance write-ups that predict it already exist, spread across a dozen systems owned by different units. When no one can find those records, interpret them, or confirm they are cleared to combine them, the shortfall surfaces as a grounded aircraft or a convoy that cannot move. The data was there. The commander simply could not reach it in time to act.
A data catalog is an organized inventory of an organization's data assets that records each asset's metadata, location, meaning, lineage, and access rules so people and systems can discover, understand, trust, and reuse data without first owning it. It is the index and governance layer over data that stays in its own systems, which is why a command can use it without first moving everything into one place.
Illustrative scenario, composite and unclassified.
A brigade sustainment officer is building the morning logistics status (LOGSTAT) roll-up ahead of a planned push. Fuel figures arrive in a theater logistics system. Ammunition on hand sits in a separate property book. Maintenance status for the brigade's vehicles lives in a third system, and a coalition partner's fuel point reports in a spreadsheet emailed over a partner network. Each source is accurate. None of them knows the others exist.
Without a catalog, the roll-up is a manual hunt. A staff NCO knows two of the sources from experience, re-keys their numbers by hand, and calls around for the rest. The partner's figures arrive late because no one can tell at a glance which fields are releasable. The commander gets a roll-up that is hours old and missing a source, and the gap in fuel capacity on the planned route does not show until the convoy is already committed.
With a catalog, the same officer searches for "fuel on hand, Class III, sector east" and gets every registered source that matches by meaning, not by file name. Each result shows its owner, how current it is, its lineage back to the reporting unit, and its classification and releasability markings. The officer can see which partner fields are releasable before asking. The roll-up is built from known, current, cleared sources, and the fuel gap shows up while there is still time to reroute.
The catalog did not move or copy any of that data. It made the data findable, interpretable, and trustworthy at the moment of decision.
A catalog centralizes knowledge about data, not data
A catalog registers what exists, describes what it means, records classification and access, and traces lineage, so a person or a system can find data by meaning and trust it before relying on it.
Visible, accessible, and understandable, the first three VAULTIS goals in the DoD Data Strategy (2020), are exactly what a data catalog delivers.
It has to, or it fails the units that need it most. A catalog that only answers queries when the enterprise network is up is an enterprise data-office tool. Forward units operate in denied, disrupted, intermittent, and limited-bandwidth (DDIL) conditions, and that is exactly when they need to know what data they hold locally and whether they can trust it. The 2023 DoD Data, Analytics, and Artificial Intelligence Adoption Strategy makes the same point: some users have abundant data and processing, while "other users face limited bandwidth at the tactical edge," and designers must identify those constraints early.
An edge-capable catalog should do four things:
Torch.AI's data-availability layer, ORCUS, is built for this environment. ORCUS connects to data where it already lives, synchronizes multi-source data, captures metadata and provenance through its Apache NiFi ingestion pipelines, enforces identity and access controls, and runs in cloud, on-premise, and disconnected tactical edge environments. It has been deployed on edge hardware as small as AWS Snowball devices and Raspberry Pi nodes, so the same governed ingestion and metadata capture that runs in the enterprise also runs forward.
The Department of War (DoW), formerly the Department of Defense, runs on data held in systems built at different times, by different organizations, under different schemas. GAO's 2025 review of command and control found that the department "largely relies on inefficient 'swivel chair' data collection and assessment processes," in which analysts "collect, assess, and package the data by manually entering and transferring them into yet another system" (GAO-25-106454). Analysts cannot connect sources they cannot find.
Current policy puts data at the base of everything else. The 2023 DoD Data, Analytics, and Artificial Intelligence Adoption Strategy describes an AI Hierarchy of Needs as "a pyramid with quality data as its foundation since all analytic and AI capabilities require trusted, high-quality data." It ties that foundation to decision-advantage outcomes that include battlespace awareness and understanding; fast, precise, and resilient kill chains; and resilient sustainment support. Its quality-data appendix asks directly whether users can "discover and rapidly identify who is responsible for specific data products, the location of data products, the types of data products available, and the means of accessing the data products," and whether organizations use "data catalogs, and metadata tagging." Those are catalog questions. The appendix builds on the VAULTIS goals first set in the 2020 DoD Data Strategy: data that is visible, accessible, understandable, linked, trustworthy, interoperable, and secure.
Torch.AI builds government-owned reasoning infrastructure, and the catalog function runs through all three core products:
Where it runs today. ORCUS is deployed for a U.S. Navy program across IL4/IL5, IL6, and the TS/SCI (JWICS) environment, in cloud, on-premise, and edge configurations. ORCUS holds three-year Authorizations to Operate (ATOs) on its IL6 and TS/SCI deployments and is operating on JWICS. The same government-owned codebase is deployed for U.S. Army programs across IL2, IL4, and IL5.
The government keeps control of the metadata and access logic, rather than renting them inside a vendor platform. See how this is packaged on the Torch.AI software page.
Evaluating a capability? Score candidates against these eight requirements. Bring them to a scoping call and we will walk each one against your environment: request a technical walkthrough.
A data dictionary defines the fields inside one dataset. A data catalog inventories whole datasets across the enterprise and governs access to them; a dictionary is usually one facet of the metadata a catalog holds.
| Term | What it covers | Relationship to a data catalog |
|---|---|---|
| Data dictionary | Fields, types, and definitions inside one dataset | One component of catalog metadata |
| Metadata management | The practice of capturing and maintaining data about data | The discipline a catalog operationalizes |
| Data governance | Policies for ownership, quality, classification, and access | The rule set a catalog enforces at discovery and access time |
| System of record | The authoritative system that holds the data | The source a catalog points to, without replacing it |
| Data lineage | Where data came from and how it changed | A metadata facet the catalog captures and exposes |
It registers data assets, records their meaning, lineage, and classification, and lets people and systems find the right data by meaning across many systems.
Because mission data is split across systems that manage it differently, which keeps analysts re-keying data by hand. The 2023 adoption strategy puts quality, discoverable data at the base of every analytic and AI capability.
It should. An edge-capable catalog keeps metadata at the node, supports local discovery when disconnected, and reconciles with the enterprise when the link returns.
A dictionary defines the fields in one dataset. A catalog inventories datasets across the enterprise and governs access to them.
No. It stores metadata and a pointer. The data stays in its system of record under that system's classification and access controls.
Describe the problem.
Define the constraints. We will structure the session.
No demo.
No slide deck.
Real work.
A data catalog makes data findable, governed, and reusable. Describe your systems, and we will scope a session around them. No new repository. Your mission data. Real work.
