Zalando’s Delta Sharing Pilot: A Governed Path for Partner Data
Zalando documents a first implementation of Delta Sharing for partner insights, including governance and manual onboarding limits.
On this page

Retail partners often need shared operational data [1] , but recurring file exports create work on both sides. A sender must prepare and deliver files; recipients then ingest them, reconcile schemas and track whether a new snapshot has arrived. An open data-sharing protocol offers another option: grant a recipient access to a table through a defined interface.
Zalando Engineering describes evaluating Delta Sharing for direct partner data access. The post presents a first implementation and its integration with the company’s data governance environment. It also describes limits: partner onboarding still involved manual steps. This is a useful pilot report, not evidence that Zalando had completed a global, self-service partner platform.
The protocol boundary
Delta Sharing defines a REST-based way [2] for recipients to list shared tables and request data. Depending on the sharing server and storage integration, the response can include metadata and temporary URLs or credentials that let the recipient read underlying files. The protocol separates the sharing interface from a specific client engine.
This can reduce the need to create and manage a separate full export for every consumer. It does not mean no bytes move: a recipient must still read the data over a network, and cloud egress, storage layout and query behaviour remain relevant. “Zero-copy” is therefore too broad as a guarantee. The practical question is whether a given deployment avoids maintaining redundant copies while meeting the consumer’s performance and isolation requirements.
Governance meets onboarding
Zalando’s article describes using its existing governance setup as part of the sharing solution. The architecture must bind a partner identity to the permitted data and maintain clear ownership over what is exposed. The exact enforcement model depends on the server and catalog integration; the open protocol itself does not automatically provide every row-level policy or audit workflow.
The operational gap was onboarding. Creating recipients and supplying credentials required manual work in the initial implementation. That can be acceptable for a pilot, but it constrains scale and increases the chance of mistakes as partner numbers rise. Credential rotation, revocation and support procedures are part of the product, not incidental paperwork.
What teams can learn from the pilot
A data-sharing protocol can create a more consistent interface than bespoke scheduled exports, particularly when multiple client tools need access. But it should be evaluated end to end: who authorizes a recipient, how access is revoked, what data files are reachable, how usage is audited, and which network charges apply.
Zalando’s case study is valuable precisely because it reports both the architectural direction and an implementation limitation. Treat partner sharing as an ongoing capability with clear security and onboarding ownership. Validate those controls in the actual catalog and storage integration before promising self-service or “zero-copy” delivery.
Validate the actual access path
A protocol-level description is not a complete security review. An implementation must verify that the recipient can access only the intended share, that temporary file access expires, and that revoking a recipient invalidates future metadata and data requests. Storage credentials and URL lifetimes should be treated as part of the authorization boundary, not as harmless transport details.
The consumer experience also depends on client support and table semantics. A partner may need compatible readers, a way to track versions and a process for handling schema evolution. For large tables, partition layout and file sizes influence query cost even if the transfer mechanism is convenient. Before offering a share broadly, test representative queries from the recipient environment and make the operational owner clear on both sides.
Zalando’s pilot makes a useful distinction between adopting an open protocol and finishing the surrounding product. The protocol can standardize requests, but recipient provisioning, credential lifecycle, support and audit are implementation responsibilities. Documenting the manual steps is a valuable signal: it tells other teams where the next engineering investment is required, rather than suggesting that “open” means “automatic.”
Frequently asked questions
Was Zalando’s article a report of a fully automated production rollout?
No. Zalando describes an evaluation and first implementation of Delta Sharing for partner insights. Its post discusses manual onboarding and open questions, so it should be understood as a pilot rather than a finished automated platform.
What is Delta Sharing?
Delta Sharing is an open protocol for sharing access to tabular data. Depending on the server and storage configuration, a recipient can obtain metadata and temporary access to data files without requiring a conventional full export into a separate warehouse.
Does Delta Sharing guarantee zero-copy or eliminate data transfer?
No. A recipient still reads data, and the sharing server’s implementation determines how access is granted. The protocol can avoid a mandatory duplicate export, but it does not eliminate network transfer, storage or egress costs.
What onboarding limitation did Zalando identify?
Zalando’s article says partner setup was manual in its first implementation, including credential handling. The post identifies smoother recipient onboarding as work to address rather than claiming self-service was already complete.



/* Comments */
Comments are offline right now — we reconnect automatically, nothing is lost.