---
title: "Workday Data Cloud Zero-Copy for Snowflake & Databricks"
description: "Configure Workday Data Cloud zero-copy sharing across Snowflake, Databricks, and Google Cloud BigLake using Apache Iceberg catalogs and OAuth federation."
url: https://kiweely.com/guides/workday-data-cloud-zero-copy-snowflake-databricks
---

# How Workday Data Cloud Zero-Copy Sharing Works with Snowflake, Databricks, and BigQuery

Workday Data Cloud shares tables with Snowflake, Databricks, and Google Cloud through Apache Iceberg REST catalogs rather than file extracts. Outbound Data Lake exposes opted-in Workday columns, while Zero-Copy Data In mounts external namespaces inside Prism Data Catalog.

-   **Checked** against Workday on November 13
-   Applies to **Workday Data Cloud**
-   6 min read
-   Published November 13
-   By Kiweely

Workday Data Cloud zero-copy sharing exchanges data with **Snowflake**, **Databricks**, and **Google Cloud (BigQuery / BigLake)** through **Apache Iceberg REST catalogs** rather than exporting CSV or Parquet extracts. For outbound sharing (**Workday Data Lake**), external platforms mount Workday’s Polaris catalog using OAuth 2.0 credentials tied to an Integration System User (ISU); for inbound sharing (**Zero-Copy Data In**), Workday’s **Prism Data Catalog** mounts your external Iceberg namespace using key-pair or OIDC token-exchange federation.

Because zero-copy connections broker live catalog metadata and short-lived storage credentials, every failed handshake comes down to one of three things: an incomplete key-pair or JWK registration, an unmaterialized column opt-in, or a missing Prism catalog security domain.

## How Does Bidirectional Zero-Copy Architecture Work Under the Hood?

Despite marketing umbrella terms like “Workday Data Connect,” outbound and inbound zero-copy flows use opposite catalog roles and completely separate Workday security domains:

-   **Outbound (Workday Data Lake -> Snowflake, BigQuery, Databricks):** Workday acts as the Iceberg catalog host (using a Polaris catalog endpoint with warehouse name `workday`, realm set to your tenant name, and principal role `all-reads-role`). Nothing in the **Data Cloud Catalog** is materialized by default. You opt in core tables, explicitly check **Include** on required columns, set a refresh schedule, and bind an [ISU and API Client](https://kiweely.com/guides/workday-integration-system-user-api-client) to a **Data Lake Access Role**.
-   **Inbound (Snowflake, Databricks, Google Cloud BigLake -> Workday Zero-Copy Data In):** Your external warehouse acts as the Iceberg REST catalog host, and Workday acts as the federated query client inside **Prism Data Catalog**. Saving a connection in Workday generates a public certificate (`PKCS#8` or `JWK`) that you attach to the external service principal so Workday can sync table schemas without copying underlying rows.

| Platform | Outbound (Workday Data Lake) Status & Connector | Inbound (Zero-Copy Data In) Catalog Endpoint Pattern | Inbound Auth & Certificate Format |
| --- | --- | --- | --- |
|  **Snowflake** | GA via **Workday Data Cloud Zerocopy Connector** (catalog-linked database) | `https://<account_locator>.snowflakecomputing.com/polaris/api/catalog` (lowercase only) |  **OAuth 2.0 with Key Pair**; download **PKCS#8** RSA public key and append fingerprint to Issuer |
|  **Databricks** |  **Workday Data Connect** query federation in Unity Catalog (Early Access; enable via Workday AE) | `<Databricks_Warehouse_URL>/api/2.1/unity-catalog/iceberg-rest` |  **OAuth 2.0 with Token Exchange**; download **JWK** and append to Service Principal federation JWKS |
|  **Google Cloud (BigQuery / BigLake)** | GA via BigLake cross-cloud connection for Workday Data Lake | `https://biglake.googleapis.com/iceberg/v1/restcatalog` with `gs://` warehouse name |  **OAuth 2.0 with Token Exchange**; OIDC **Workload Identity Pool** + upload **JWK** key |

## How Do You Configure Outbound Workday Data Lake Without Empty Tables?

The most common outbound mistake is registering an API Client and mounting the `workday` catalog in Snowflake or BigQuery before materializing columns or running an initial refresh. Follow the exact seven-part tenant sequence:

1.  **Enable the Data Cloud functional area:** Run **Maintain Functional Areas**, check **Enabled** on *Data Cloud*, and run **Activate Pending Security Policy Changes**.
2.  **Grant Data Lake admin and provisioning domains:** Configure permissions on *Data Cloud: Manage Data Lake Application Configurations*, *Data Cloud: Manage Data Lake Access Policies*, and *Manage: Datalake User Provisioning*, then activate security changes.
3.  **Create a dedicated ISU and API Client:** Create an Integration System User for Data Lake and run **Register API Client** using either the **JWT Bearer Grant** (associated with an RSA 2048 `.pem` key) or **Refresh Token Grant**, depending on your external platform connector.
4.  **Opt in core tables and individual columns:** Open the **Data Lake Admin Console** report, go to the **Applications** tab, and select **Edit** on *Workday Core*. Select core tables in **Tables for Data Lake Materialization**, then open **Manage Fields** for each table. Aside from ID columns, every column is excluded by default—check **Include** on each required column, and opt in any associated tables that appear in the **Opt-In Reference Tables** grid for multi-instance array fields.
5.  **Set the materialization schedule and run at least once:** On the **Schedule** tab, set **Run Frequency** (Production supports hourly on Data Cloud Pro and daily on Core; Sandbox, Preview, and Implementation tenants support weekly). Wait for the first refresh to complete—until it runs once, the **Object Access** prompt for access roles stays empty.
6.  **Bind the ISU to a Data Lake Access Role:** On the **Access Management** tab of **Data Lake Admin Console**, select **Add Data Lake Access Role**, pick your Data Lake ISU, and select the exact tables that principal may read.

Joiner Tables for Engines Without Array Unnesting

Workday stores multi-instance relationships as `ARRAY(BIGINT)` foreign keys. If a downstream tool or federated view cannot execute `UNNEST`, query the pre-flattened bridge tables in `workday.join_workday_core` using the naming pattern `join_<source_table>_to_<target_table>_via_<source_fk_column>`.

## How Do You Configure Zero-Copy Data In for Snowflake, Databricks, and BigQuery?

Before opening **Prism Data Catalog**, configure Workday security prerequisites: run **Maintain Assignable Roles** to enable the *External Table Owner*, *External Table Editor*, and *External Table View* roles for *Prism External Table*, grant your admin security groups permissions on *Prism: Manage Connection*, *External Catalog Create*, *External Catalog Manage*, *External Catalog Owner Manage*, *External Tables Manage*, and *External Tables Owner Manage*, and run **Activate Pending Security Policy Changes** (log out and back in so `Prism: Manage Connection` takes effect).

Next, in **Prism Data Catalog**, select **\+ Create > Connection** under **Zero Copy Connectors**. Save the connection **without** clicking **Test Connection** first—you must complete the certificate round-trip for your specific cloud platform:

### Snowflake Key-Pair Fingerprint Round-Trip

Snowflake requires lowercase catalog URLs and a three-part `Issuer` string containing the public key fingerprint that Snowflake calculates *after* you upload Workday’s PKCS#8 certificate:

Fails

```
-- Saving the initial placeholder Issuer in Workday and clicking Test Connection immediately:
-- Issuer in Workday: XTQ04827.ICEFLOW_PRISM_USER
-- Fails OAuth key-pair validation because RSA_PUBLIC_KEY is unset and fingerprint is missing.
```

Works

```
-- 1. Download Certificate > PKCS#8 from Workday Connection Details, then run in Snowflake:
ALTER USER ICEFLOW_PRISM_USER SET RSA_PUBLIC_KEY = 'MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8A...';
DESCRIBE USER ICEFLOW_PRISM_USER;

-- 2. Copy RSA_PUBLIC_KEY_FP (e.g., SHA256:abc123xyz...) and edit the Workday Connection Issuer to:
-- XTQ04827.ICEFLOW_PRISM_USER.SHA256:abc123xyz...
-- 3. Save in Workday, then click Test Connection.
```

### Databricks Unity Catalog JWK Federation

1.  Enter Catalog URL `<Databricks_Warehouse_URL>/api/2.1/unity-catalog/iceberg-rest`, Authentication Type **OAuth 2.0 with Token Exchange**, Issuer `https://databricks.com/account/<account_id>/oidc/v1/token`, and Endpoint `<Databricks_Warehouse_URL>/oidc/v1/token`.
2.  Save the connection in Workday, then select **Certificate > Download > JWK**.
3.  In the Databricks account console, open your Service Principal’s federation policy, append the downloaded Workday JWK entry to the `"keys"` array in the JWKS JSON, save, and return to Workday to click **Test Connection**.

### Google Cloud BigLake Workload Identity Federation

1.  In Google Cloud Console, create a **Workload Identity Pool** with an **OIDC** provider (never SAML or AWS), grant your BigLake service account read permissions on the catalog and GCS bucket, link the service account to the pool, and download the JSON client library config to get the exact `audience` and `token_url` values.
2.  In Workday **Create Connection > Google Cloud**, enter Catalog URL `https://biglake.googleapis.com/iceberg/v1/restcatalog`, Catalog Name (`gs://...` warehouse property), Google Cloud `Project ID`, **OAuth 2.0 with Token Exchange**, Scope `https://www.googleapis.com/auth/cloud-platform`, and the exact `Audience` and OIDC `Issuer`.
3.  Save, click **Certificate > Download > JWK**, upload a combined JWK file (preserving existing keys plus the new Workday JWK) onto your Google Cloud Workload Identity Pool provider, and then run **Test Connection** in Workday.

## How to Sync Schemas, Blend Data, and Verify End-to-End

Once **Test Connection** succeeds, complete three steps in **Prism Data Catalog** to prove zero-copy queries work:

1.  Select **\+ Create > External Catalog**, choose your connection, and enter the exact case-sensitive Iceberg **Namespace** (such as `enterprise.finance.procurement`). Saving automatically triggers a **Schema Sync** that registers external tables without copying rows.
2.  Open **External Tables**, double-click a synced table to inspect column types and sample preview values, and select **Quick Action > Edit External Table > Enable for analysis**.
3.  If blending with Workday records, select **\+ Create > Derived Dataset**, add a **Join** stage matching compatible data types (avoid joining string Worker IDs to numeric warehouse keys), and click **Quick Actions > Publish**. Confirm the status reaches *Published* before building custom reports, Discovery Boards, or [WQL queries](https://kiweely.com/guides/workday-wql).

Setting up all six Prism external catalog domains, assignable roles, and ISU security groups across Sandbox and Production is tedious by hand; [Kiweely Vine](https://kiweely.com/vine) audits missing security policies in your live tenant and prepares the exact domain and role configuration for your approval ([explore admin and consultant jobs](https://kiweely.com/vine/jobs)).

## Frequently asked questions

### Why are my Workday Data Lake tables empty or missing columns after I enable a table?

Workday Data Lake uses a strict opt-in materialization model. Adding a core table only registers table metadata and ID columns; every other column remains excluded until you open Manage Fields for Data Lake Materialization, select Include, and let the scheduled refresh run at least once.

### Why does Test Connection fail immediately when setting up Zero-Copy Data In?

Do not click Test Connection right after saving the connection in Prism Data Catalog. You must first download the Workday-generated certificate (PKCS#8 for Snowflake, JWK for Databricks and Google Cloud) and register that public key on the external platform's service user or federation policy before testing.

### How does Google BigQuery connect to Workday Data Cloud zero-copy sharing?

Google Cloud connects via BigLake Iceberg REST catalogs and cross-cloud lakehouse connections. For Zero-Copy Data In, Workday authenticates to https://biglake.googleapis.com/iceberg/v1/restcatalog using OAuth 2.0 with Token Exchange backed by Google Cloud Workload Identity Federation (OIDC).

### Are new Workday release fields automatically added to existing Data Lake shares?

No. Tables and columns introduced in future Workday releases are never materialized automatically. Administrators must explicitly opt in new fields in the Data Lake Admin Console, or migrate updated configurations via Object Transporter.

## Sources checked

-   [Get Started with Workday Data Lake](https://doc.workday.com/admin-guide/en-us/workday-data-cloud/data-out/workday-data-lake/workday-data-lake.html)
-   [Add Tables and Columns to Data Lake](https://doc.workday.com/admin-guide/en-us/workday-data-cloud/data-out/workday-data-lake/add-tables-and-fields-to-data-lake.html)
-   [Set Up Zero-Copy Data In with Snowflake](https://doc.workday.com/admin-guide/en-us/workday-data-cloud/data-in/set-up-zero-copy-data-in-with-snowflake.html)
-   [Set Up Zero-Copy Data In with Databricks](https://doc.workday.com/admin-guide/en-us/workday-data-cloud/data-in/set-up-zero-copy-data-in-with-databricks-2.html)
-   [Set Up Zero-Copy Data In with Google Cloud](https://doc.workday.com/admin-guide/en-us/workday-data-cloud/data-in/set-up-zero-copy-data-in-with-google-cloud.html)

## Related guides

[All guides](https://kiweely.com/guides)

-   [### Workday Data Cloud vs. EIB, RaaS, and Prism: Which Data Path Should You Use? Use Workday Data Cloud when Snowflake, Databricks, or BigQuery need zero-copy Iceberg or live SQL access to Workday tables without custom extract pipelines. Keep EIB and RaaS for vendor file deliveries, calculated-field payloads, and custom objects not yet in the Data Cloud Catalog.](https://kiweely.com/guides/workday-data-cloud-vs-eib-raas-prism)
-   [### Workday Prism Analytics vs. WQL vs. Custom Reports: Joining External and Tenant Data Prism Analytics blends external and Workday data into published Prism data sources, while WQL and Custom Reports query live tenant business objects. Here is how to choose between them, how Prism’s five-phase security model works, and what limits apply to Prism reports and calculated fields.](https://kiweely.com/guides/workday-prism-analytics-vs-wql)
-   [### How to Budget Workday Flex Credits Before You Sign Flex Credits are drawn down per metered action, so a budget starts from transaction volumes, not headcount. This guide walks through the estimate, the sandbox pilot and the contract questions.](https://kiweely.com/guides/workday-flex-credits-budget-planning)
-   [### How do I set up an integration system user and API client in Workday so an external app can authenticate? Create the ISU, put it in an Integration System Security Group, grant domain permissions and activate them, register the client for integrations, then generate its refresh token and pick the ISU as the Workday account.](https://kiweely.com/guides/workday-integration-system-user-api-client)

### More guides on More Workday guides

[All 46 guides](https://kiweely.com/guides#more)

-   [Workday Platform Consumption Console: what each page shows, when usage updates and what the alerts do](https://kiweely.com/guides/workday-platform-consumption-console)
-   [Workday Tickets for a Small HRIS Team: Workday Help, Jira or ServiceNow, and How to Triage the Queue](https://kiweely.com/guides/workday-hris-ticketing-help-jira-servicenow)
-   [What Workday AMS Costs in 2026: Hourly Rates, Retainers and What They Cover](https://kiweely.com/guides/workday-ams-cost)
-   [Hire Another Workday HRIS Analyst or Add an AI Team? The Monthly Cost Compared for 2026](https://kiweely.com/guides/hris-analyst-or-ai-team-cost)
-   [Workday Configuration Catalog packages: filter, test, then migrate](https://kiweely.com/guides/workday-configuration-catalog-packages)
-   [How do I limit a Configuration Change Tracker package to a change window?](https://kiweely.com/guides/workday-configuration-change-tracker-time-bounded-packages)
-   [How do I build a Configuration Change Tracker package by user and time?](https://kiweely.com/guides/workday-configuration-change-tracker-how-to)
-   [Do You Need a Workday AMS Partner, or Can a Small In-House Team Run Workday?](https://kiweely.com/guides/workday-after-go-live-in-house-team)

### From the blog on More Workday guides

[All posts](https://kiweely.com/blog)

-   [How Sana Concierge and Workday Everywhere Change HR Tier-1 Support Workflows](https://kiweely.com/blog/workday-sana-concierge-hr-service-delivery)
-   [Everyone Knows AI Can’t Read Workday Community. So We Fixed It.](https://kiweely.com/blog/ai-cant-read-workday-community)
-   [The AGI Moment for the Workday Ecosystem](https://kiweely.com/blog/agi-moment-workday-ecosystem)

## Let Workday check the work.

Kiweely writes WQL and builds orchestrations in your own tenant, and Workday, not Kiweely, says whether they passed.

[Start for free](https://kiweely.com/download) [See how it works](https://kiweely.com/vine/how-it-works)

## Get the next guide in your inbox.

One email a month with new guides and one Extend tip. Nothing else.
