Sharing data with a client, a partner or even with another department of your own organization usually ends the same way: someone exports a file, uploads it to a server, sends it by email, and by the time it reaches the recipient it’s already out of date. Every copy multiplies the security risk, the storage cost and the doubt about which version is the right one. And if the two parties work on different platforms, the problem gets worse.
Delta Sharing exists precisely to solve that. It’s an open protocol that gives access to live data, between organizations and across platforms, without copying or replicating it. In this article you’ll see what Delta Sharing is, how it works, and how it combines with cross-format interoperability so that the same dataset can be read as Delta Lake or as Apache Iceberg without duplicating anything.
The aim is that, as a data or digital transformation lead, you understand when it makes sense to adopt it and what architecture decisions it involves.
What Delta Sharing is
Delta Sharing is an open protocol, developed by Databricks and donated to the Linux Foundation, for securely sharing data with other organizations, regardless of the platform or cloud each party uses. The key is that openness: the recipient doesn’t need an account with your same provider or to install your platform. They only need a client compatible with the protocol, and there are connectors for tools as common as Power BI, Apache Spark, pandas and Tableau.
The underlying principle is simple to explain and powerful in practice: you share access, not files. Instead of generating a copy that travels to the recipient, Delta Sharing grants read access to the data that already lives in your cloud storage. The recipient always queries the most recent version, and you keep control at all times over what is shared and with whom.
How Delta Sharing works
The model is organized around three elements: the provider (who offers the data), the share (the grouping of tables you decide to expose) and the recipient (the organization you grant access to). You define which tables, views or even specific partitions go into each share, and which recipients you authorize. That granularity matters: you can share only part of a table, or apply filters so each recipient sees only the rows that belong to them.
Technically, the exchange relies on a simple REST API. When a recipient requests data, the server verifies their permissions, logs the request and generates short-lived pre-signed URLs that let them read the files directly from cloud storage. This way the transfer is parallel and uses the bandwidth of the cloud system itself, without the data having to pass through an intermediate server.
For large datasets, this design is fast and inexpensive, and it also allows sharing across different clouds and regions without replicating the information.
Sharing data across formats: Delta Lake and Iceberg without duplication
Sharing between organizations solves half the problem. The other half appears when the two parties work with different table formats. For years, companies have had to choose between Delta Lake and Apache Iceberg as their storage standard, knowing that each query engine and each platform supported one or the other better. Standardizing on a single format always meant giving up part of the ecosystem.
The good news is that this choice is no longer mutually exclusive. Both Delta Lake and Iceberg are built on the same data files in Apache Parquet format; what changes, fundamentally, is the metadata layer. Taking advantage of that overlap, today it’s possible to keep a single copy of the data and expose it in both formats at once.
Delta UniForm and the Iceberg REST catalog
The mechanism that makes this possible is called Delta UniForm (Delta Lake Universal Format). When you enable it on a Delta table, the system automatically generates the equivalent Iceberg metadata alongside the Delta metadata, without rewriting or duplicating the underlying Parquet files. The result is that an Iceberg client can read a table written as Delta Lake, accessing a single physical copy of the data.
On top of this, Unity Catalog implements the Iceberg REST catalog, the standardized interface the Iceberg community uses to discover and read tables. Thanks to it, external engines and platforms such as Trino, Snowflake or DuckDB can access your data as if they were native Iceberg tables. Databricks’ commitment to this convergence was reinforced by the acquisition of Tabular (the company founded by Iceberg’s creators) in 2024, and by subsequent progress in catalog unification. The practical consequence for your organization is that cross-format interoperability stops being a migration project and becomes a configuration.
Governance and security: the role of Unity Catalog
Sharing data openly is only viable if you can govern it with the same rigor you apply internally. This is where Unity Catalog acts as the central layer: it integrates Delta Sharing natively, so that managing providers, shares and recipients happens in the same place where you define permissions, auditing and traceability for the rest of your data assets.
This matters especially in regulated sectors such as pharma, finance or industry, where demonstrating who accessed what data, when and with what permissions isn’t optional. With Unity Catalog’s centralized auditing, every access to the shared data is logged, and you can revoke permissions or adjust the scope of a share at any time. Sharing stops being a blind spot in your data governance strategy and becomes part of it.
Use cases and benefits for your company
Beyond the mechanics, it’s worth pinning down where this approach adds real value. These are the scenarios where Delta Sharing and cross-format interoperability make a tangible difference for medium and large companies.
- The first is collaboration with clients and partners. A company that distributes data to multiple consumers (each with their own tools) can publish a single share and let each recipient consume it from their preferred platform, always with the most recent version. Manual transfers and out-of-date copies disappear.
- The second is data mesh architecture within a single organization. Different business units or subsidiaries, which often operate in different clouds or regions, can share data with each other without replicating it, while keeping common governance.
- The third is multi-platform analytics. Teams working with different engines access the same data without anyone having to duplicate or convert it.
In every case, the benefits converge on four points: eliminating copies and, with them, reducing storage cost; access to live data instead of stale snapshots; less dependence on a single vendor, by relying on open formats and protocols; and a contained transfer cost, since Delta Sharing makes it possible to share even with zero egress fees using compatible storage. For a leadership team assessing the return on its data platform, these are concrete, measurable arguments.
How we approach data sharing and interoperability at BertIA
At BertIA, we treat data sharing as an architecture decision, not as an isolated tool. Before enabling Delta Sharing or UniForm, we assess how your data model is organized, which assets make sense to expose, to whom, and with what governance controls. That prior diagnosis avoids the most common mistake: opening access to data that isn’t yet sufficiently governed.
Our work combines implementation on Azure Databricks and Unity Catalog with sector knowledge of the environments we operate in, where traceability and compliance shape every technical decision.
We define which formats are worth exposing according to the real consumers of each piece of data, configure the interoperability so your information is readable from the analytics ecosystem you already use, and leave the auditing ready to withstand a regulatory review. The goal isn’t to share more, but to share the right thing, with the right people, in a controlled way.
Conclusion
Delta Sharing changes the way you think about data exchange: instead of moving copies that age and multiply risks, you grant governed access to live data, across platforms and across clouds. Combined with the cross-format interoperability that Delta UniForm and the Iceberg REST catalog provide, your organization can keep a single copy of the data and let each consumer read it with their own tools, with no migrations or duplications.
The question stops being “which format do I choose” and becomes “how do I properly govern what I share.”
If you want to design a data sharing and interoperability strategy tailored to your architecture and your sector, talk to our team of experts and we’ll assess the starting point and the next steps together.
