Independent IFS Cloud practice · Data mesh series, step 7
Data discoverability: let every team find a data product, its owner and its meaning without asking
Data discoverability means that a consumer can find a data product, see who owns it, understand what it contains and learn how to get access, all without sending an email. In a data mesh it is a required quality of every data product, and it depends on automatic registration in a catalog, not on a better spreadsheet.
Key takeaways
- Discoverable is the first quality of a data product in Zhamak Dehghani’s 2019 definition, and her article names a central catalog where each domain registers its products as the common way to achieve it.
- A catalog entry needs technical metadata (interface, schema), business metadata (owner, definitions, service levels) and usage metadata (who uses the product and how often).
- For data served from IFS Cloud, the API Explorer provides the OpenAPI and OData specification of each projection, which a catalog can register instead of a hand-written schema.
- Registration belongs in the delivery pipeline: a product without an owner or a description is not promoted to the shared zone.
- Finding a product is not the same as reading it. The catalog shows that data exists, and access is still granted separately, in IFS Cloud through a permission set.
1.What is data discoverability in a data mesh?
Data discoverability is the property that lets a consumer locate a data product and judge whether to use it from its catalog entry alone. In Dehghani’s original data mesh article (martinfowler.com, May 2019), discoverable is the first of the basic qualities a domain data product must have, and the article describes a registry or catalog holding owners, origin, lineage and sample datasets as the common implementation.
Decentralized ownership makes discoverability harder, not easier. When one central team held all data, people at least knew whom to ask. Once ownership moves to the domains, as described in the data mesh overview and its eight steps, a product that is not registered is invisible, and the usual result is a second team rebuilding a dataset that already exists under a different name.
This is step 7 of the series. It builds on step 4, where each domain defines and delivers its data products with metadata and service levels: discoverability makes that metadata searchable.
2.What metadata does a data product need to be found and trusted?
A catalog entry needs enough metadata for a consumer to decide without contacting the owner, and every field should come from a source that keeps it current. The table lists a practical minimum for a product served from IFS Cloud.
| Field | Type | Where the value comes from |
|---|---|---|
| Product name and description | Business | The domain team, written for a reader outside the domain |
| Owner | Business | The named data product owner, as a role and a person |
| Interface | Technical | The projection or other endpoint the product is served through |
| Schema | Technical | The OpenAPI or OData specification, taken from the source system, not retyped |
| Business definitions | Business | The shared glossary, so that a term means the same in every domain |
| Service levels | Business | The product specification: refresh frequency, accuracy commitment, support contact |
| Lineage | Technical | The platform’s pipelines, showing which sources and transformations feed the product |
| Access route | Governance | The name of the permission set or role a consumer requests |
| Usage | Usage | The catalog itself: which teams use the product and how often |
Usage metadata matters to consumers because a product that other teams already depend on is easier to trust than one nobody has opened. The wider practice behind these fields is covered in metadata management.
3.How do you describe IFS Cloud data in a catalog?
IFS Cloud data is described in a catalog through the projections that serve it, because each projection is an OData service with a machine-readable specification. The API Explorer gives an overview of the projections in a system, including the Premium APIs designed for integration, and provides OpenAPI v2, OpenAPI v3 and OData specifications for each one.
Two practical points follow. First, the domain team names the exact projection behind a product. When a key user knows the IFS Cloud Web page where the data lives, the Debug Console on that page shows page info including the projection names, as the API Explorer documentation describes, which is faster than searching the full list. Second, the schema in the catalog is generated from the specification, so it changes when the projection changes in a release update, instead of drifting away from the system as a hand-maintained description would.
Catalog tools such as Alation, Collibra or Amundsen can hold these entries next to products from data warehouses and lakes. The choice of tool matters less than whether registration is automatic. Before choosing, check how the tool reads OpenAPI or OData specifications and how it handles the authentication your IFS Cloud environment requires. Mapping domains to projections and setting up registration is part of my services for IFS Cloud.
4.Why must catalog registration be automatic?
Catalog registration must be automatic because a catalog filled by hand goes stale from the first release after it is filled. Owners stop updating entries once the initial effort is over, and a catalog with stale entries is worse than none, because consumers trust it.
The working pattern is to register products from the same CI/CD pipeline that deploys them. The pipeline reads the product’s metadata from a file kept with its code, pushes it to the catalog and refuses to promote the product to the shared or certified zone when a required field, such as the owner or the description, is empty. That check is federated computational governance applied to discoverability: a rule agreed once and enforced by the platform, as described in apply federated computational governance.
Lineage is collected the same way. When a consumer sees a wrong figure in a report, lineage lets them follow it back to the transformation or the source that produced it, and the question goes to the right owner instead of to a general support queue.
5.Why does a shared glossary decide whether search works?
A catalog search returns fragmented results when domains use different words for the same thing or the same word for different things. If marketing registers a prospect as “lead” and sales registers the same party as “opportunity”, a consumer searching for one finds half of the data.
The fix is a shared glossary in which each business term has one agreed definition, and a rule that every catalog entry links its fields to glossary terms. Deciding the definition is a governance task, not a catalog feature: the standing rules belong in the enterprise book of rules, and the catalog points to them.
6.Does discoverability mean everyone can read the data?
Discoverability does not grant access: a consumer sees from the catalog that a product exists and what it contains, and then requests access through the route the entry names. Showing metadata widely while keeping records protected is what lets a company make its data visible without making it public.
For a product served from IFS Cloud, the access route is a projection grant. Permission sets are created under Solution Manager > Access Control > Permission Sets and grant access to projections, among other objects. Naming that permission set in the catalog entry turns an access request into a known, reviewable change.
7.In what order should you roll out discoverability?
Discoverability is rolled out domain by domain, because indexing the whole company at once produces a large catalog with thin entries.
- One domain, one catalog. Pick the domain with a willing owner and real consumers, list its existing silos and choose a catalog tool that can read the specifications your sources provide.
- Automatic registration. Connect the delivery pipeline to the catalog and stop manual entry of technical metadata.
- Business glossary. Link catalog fields to agreed definitions and train domain teams to write product documentation for readers outside their domain.
- Usage and feedback. Show usage in the catalog and let consumers rate or comment on products, so owners see which products matter.
Each new domain repeats the first two steps against the platform the first domain proved. The people side of this, persuading owners that documentation is part of the product, is the subject of the next step: the cultural shift and training for data mesh.
8.Frequently asked questions
What is data discoverability?
Data discoverability is the ability of a consumer to find a data product, see its owner, understand its content and learn how to get access, from a catalog entry and without contacting anyone. In a data mesh, discoverable is one of the basic qualities every data product must have.
Why is a data catalog better than a shared spreadsheet of datasets?
A data catalog is updated automatically from the pipelines and source systems, so its schemas, lineage and usage stay current. A shared spreadsheet of datasets is correct only on the day someone edits it, and it cannot show lineage or usage. For data mesh, where many domains publish products, only automatic registration keeps the list complete.
How do you register IFS Cloud data in a data catalog?
IFS Cloud data is registered in a data catalog through the projections that serve it. The IFS Cloud API Explorer lists the projections in a system and provides OpenAPI v2, OpenAPI v3 and OData specifications for each, which a catalog can use as the schema. The domain team adds the owner, business definitions and service levels.
Does a data catalog give everyone access to the data it lists?
A data catalog shows metadata, not records. Consumers see that a data product exists and what it contains, and they request access separately. For a product served from IFS Cloud, that access is a projection grant in a permission set, which the catalog entry can name.
Can AI fill in a data catalog?
Catalog tools can suggest tags and descriptions automatically, but a suggested description is a draft. The data product owner still confirms the business meaning, because a wrong definition in a data catalog misleads every consumer who trusts it.
9.About the author
Dariusz Myśliwiec: 18 years hands-on with IFS (Apps 7.5–10 and IFS Cloud). IFS Certified Associate Consultant. PRINCE2® 7. Based in Kraków, delivering remotely across Europe and globally through an independent practice. You talk to the consultant who does the work, not a sales layer.
Selected clients from my IFS work: Fugro · LGC · BVI Medical · Betafence (PRÆSIDIAD) · Barlinek · NGK Ceramics · Newag · Oleofarm.
IFS is a registered trademark of IFS AB; this practice is not affiliated with IFS AB.
Can your teams find IFS Cloud data without asking?
If the answer to “where is the data?” is still a person, tell me which domain hurts most. On a call we go through which projections serve it, who should own it and what its catalog entry needs. For the broader engagement, see my services for IFS Cloud.
Book a call about data discoverability Read the data mesh overview
