Analytics Hub is one of those Professional Data Engineer exam topics that looks niche until you see how often it shows up as the right answer to data-sharing scenarios. If a question mentions sharing datasets across teams, across organizations, or with outside partners, and BigQuery is in the picture, Analytics Hub is usually the service the exam wants you to pick. I want to walk through what it does, how the publisher and subscriber model works, why copy-free sharing matters, and how Private Exchange fits in.
Analytics Hub is a centralized platform for sharing BigQuery data and collaborating on it. Google launched it in 2021 to address the growing need for data exchange both inside organizations and across organizational boundaries. The product sits on top of BigQuery and makes datasets discoverable through something that looks a lot like a marketplace, where publishers post listings and subscribers pull them in.
The mental model I use when answering Professional Data Engineer questions is simple. BigQuery stores the data. Analytics Hub controls how that data gets exposed to other projects and other organizations without anyone having to copy or move bytes around. That distinction matters because the exam loves questions where the wrong answer is something like "export to Cloud Storage and grant access" or "set up a scheduled query to replicate the table." Analytics Hub exists so you do not have to do either of those things.
Every Analytics Hub interaction has two sides. On one side is the Publisher Project, which owns the source datasets in BigQuery. The publisher creates a shared dataset and exposes it through a listing inside a data exchange. The listing is the entry that subscribers see and request access to.
On the other side is the Subscriber Project. When a subscriber accepts a listing, Analytics Hub creates a linked dataset inside the subscriber's BigQuery environment. The linked dataset is a read-only pointer back to the source data in the publisher's project. The subscriber can query it like any other BigQuery dataset, join it to their own tables, and feed it into Looker, Vertex AI, Tableau, or whatever downstream tool they use. The actual rows never leave the publisher's project.
If you remember nothing else for the exam, remember this triangle: publisher dataset, listing in the exchange, linked dataset in the subscriber project. Questions often test whether you understand that the subscriber sees a linked dataset, not a copy.
The copy-free model is the architectural advantage Analytics Hub brings to the exam. When you share data the old way, you either export it, replicate it through a pipeline, or grant cross-project IAM on a raw BigQuery dataset and hope the access boundaries hold. Each approach has drawbacks. Exports go stale. Replication adds cost and lag. Raw IAM grants are clumsy when you want to share with dozens of subscribers or revoke access cleanly.
Analytics Hub solves all three. The linked dataset queries the live source, so there is no staleness. No bytes are duplicated, so storage cost lives only in the publisher's project. And access is managed through subscriptions to listings, so revoking a single subscriber is a one-click operation. When an exam scenario emphasizes freshness, no duplication, or auditable cross-org access, that is your signal to pick Analytics Hub.
Analytics Hub supports several flavors of exchange. Public exchanges include the Google-hosted catalog where you can subscribe to public datasets like weather, demographics, or COVID data. Commercial exchanges let third-party data providers sell datasets. Private exchanges are the ones you build yourself for internal or partner sharing, and they are the most exam-relevant category because they cover the secure data-sharing scenarios the Professional Data Engineer blueprint tends to ask about.
Private Exchange is the feature that gives you a secure, controlled environment for sharing sensitive data. The benefits the exam expects you to know are enhanced data privacy, controlled access to data assets, and customizable sharing policies. Practically that means you decide exactly who can see a listing, who can subscribe, and what the data governance posture looks like before anything is exposed.
Private Exchange is the right answer when the scenario involves proprietary data inside a single company, or collaboration with a small set of trusted external partners. Healthcare providers sharing patient data, financial institutions collaborating on market research, and retailers analyzing customer behavior with a partner are all canonical examples. If the scenario adds words like HIPAA, confidential, or regulated, Private Exchange is almost certainly part of the answer.
Private Exchange also operates inside VPC Service Controls perimeters, which means you can wrap the publisher and subscriber projects in a service perimeter and still share through Analytics Hub. That detail occasionally surfaces on the exam in security-flavored questions.
Billing follows the copy-free architecture. The publisher pays for storage of the source dataset because the data physically lives in their project. Subscribers pay for the compute when they run queries against the linked dataset, since BigQuery query costs always sit with the project that issues the query. That split is worth memorizing because Professional Data Engineer scenarios sometimes ask who absorbs which cost in a cross-org sharing setup.
When you see a Professional Data Engineer question, scan for these triggers:
If two or more of these show up, Analytics Hub is the answer, and Private Exchange is the variant when the data is sensitive.
My Professional Data Engineer course covers Analytics Hub alongside the rest of the BigQuery data-sharing surface, with worked exam scenarios so you can spot the publisher-subscriber pattern quickly under time pressure.
