As I mentioned in my previous blogs, SQL on Files, a capability of SAP HANA database in SAP HANA Cloud, has been continuously expanding since SAP HANA Cloud QRC 3/2024. We started with native read-only access to CSV, Parquet, and Delta tables in SAP HANA Cloud, data lake Files. We then extended this access to external object storages (Amazon S3, Azure Blob/ADLS Gen2, Google Cloud Storage) and external Delta Sharing in QRC 4/2024. Most recently, we introduced Delta table-to-HANA replication in QRC 4/2025.

With SAP HANA Cloud QRC 2/2026, I'm pleased to announce that the journey continues, and this time the spotlight is on Apache Iceberg, the industry-standard open table format that has rapidly become a cornerstone of modern data lakehouse architectures. While Iceberg support has been quietly maturing in SAP HANA Cloud over the last few quarters, this is the first blog dedicated to telling the full story end-to-end.

In a nutshell, here is what SAP HANA Cloud now offers for Apache Iceberg, cumulatively from QRC 3/2025 through QRC 2/2026:

• Direct read-only access to Apache Iceberg tables in object storage via the file adapter, including time-travel queries (QRC 3/2025).
• Snapshot replica toggling support for Apache Iceberg virtual tables (QRC 1/2026).
• Data Page V2 (DATA_PAGE_V2) support in the Parquet reader, aligning SAP HANA Cloud with the latest Parquet ecosystem (QRC 1/2026).
• Direct read-only access to Apache Iceberg REST Catalogs via the new icebergcatalog adapter. QRC 1/2026 introduced the adapter with Cloudera and Databricks running on AWS as the initially certified providers, and QRC 2/2026 extends the multi-cloud coverage to Azure and GCP for Databricks while additionally certifying Snowflake across AWS, Azure, and GCP. Cloudera certification on Azure and GCP is planned for a future release.
• AliCloud Object Storage Service (OSS) added to the list of supported external object storages (QRC 2/2026).

Alright, with these points in mind, let's dive deeper into the details with some examples.

Apache Iceberg via the file Adapter

Starting with SAP HANA Cloud QRC 3/2025, SQL on Files began supporting Apache Iceberg as an additional file format on top of the existing CSV, Parquet, and Delta table support. As highlighted in my first SQL on Files blog and follow-up blog, the file adapter provides direct read-only access to files in SAP HANA Cloud, data lake Files or in supported external object storages. With Iceberg now in the mix, customers can query Iceberg tables sitting on object storage without any data movement or ingestion.

Since Apache Iceberg is an open table format, the way of creating a virtual table is similar to Delta tables, where you only need to point the virtual table to the root of the Iceberg table:

-- create a virtual table by pointing to an apache iceberg table
CREATE VIRTUAL TABLE DEMO.ICEBERG_ORDERS AT "demo-hdlf_rs"."/iceberg/orders/" AS ICEBERG;

A few things worth knowing about this support:

• Both Iceberg v1 (Copy-on-Write) and v2 (Merge-on-Read) are supported for read-only access.
• Time-travel queries are supported, similar to Delta tables, via either snapshot ID or timestamp:

-- time travel by snapshot id
SELECT * FROM DEMO.ICEBERG_ORDERS FOR VERSION AS OF '2953155114686890267';

-- time travel by timestamp
SELECT * FROM DEMO.ICEBERG_ORDERS FOR SYSTEM_TIME AS OF '2025-04-08 07:43:31.793000000';

• Two new built-in functions help you explore Iceberg metadata directly from SAP HANA Cloud:

• The existing GET_REMOTE_SOURCE_FILE_COLUMNS built-in procedure has been extended to retrieve column information from Iceberg tables as well.

Important: Apache Iceberg metadata stores absolute table paths. Simply copying Iceberg files from one location to another will cause inconsistencies, and creating virtual tables on the copied files will not work. Always create or migrate Iceberg tables using a proper Iceberg-compatible engine.

Please refer to the links below for further details.

Apache Iceberg REST Catalog via the icebergcatalog Adapter

Reading Iceberg tables directly from object storage with the file adapter is great for self-managed scenarios. However, in real-world enterprise lakehouses, Iceberg tables are usually managed by an Iceberg REST Catalog server, for example, Databricks Unity Catalog, Cloudera Data Platform, or Snowflake. Customers told us that, for production-grade integration, the catalog-based approach is what they consider enterprise-ready: it centralizes table management, access control, and metadata governance.

That's where the new icebergcatalog adapter comes in. Starting with SAP HANA Cloud QRC 1/2026, SQL on Files supports direct read-only access to external Apache Iceberg REST Catalogs, with Cloudera and Databricks running on AWS as the initially certified providers. With QRC 2/2026, the multi-cloud coverage is extended to Azure (ADLS Gen2) and GCP (GCS) for Databricks, and Snowflake is added as an additionally certified provider across AWS, Azure, and GCP. Cloudera certification on Azure and GCP is planned for a future release.

The way to create a remote source to an Iceberg REST Catalog follows the official Apache Iceberg REST Catalog specification with bearer-token (OAuth) authentication:

-- create a remote source to an iceberg rest catalog
CREATE REMOTE SOURCE  ADAPTER "icebergcatalog" CONFIGURATION '
provider=databricks;
endpoint=;'
WITH CREDENTIAL TYPE 'OAUTH' USING 'access_token=';

Once the remote source is created, virtual tables are created in the same way as with other SAP HANA smart data access (a.k.a., SDA) adapters by pointing to the catalog → namespace → table:

-- create a virtual table on a table exposed via the iceberg rest catalog
CREATE VIRTUAL TABLE DEMO.VT_ORDERS
  AT ""."".""."";
A few important notes on the certification scope and authentication:
The icebergcatalog adapter is designed to work with any spec-compliant Apache Iceberg REST Catalog. In theory, any REST catalog that follows the official spec and bearer-token authentication should work.
However, because small differences in authentication flows or vendor-specific extensions can cause unexpected issues, SAP officially certifies only a defined set of Iceberg REST Catalog providers per QRC, as summarized below. Other REST catalogs may work but are not officially certified.
Provider
AWS
Azure (ADLS Gen2)
GCP (GCS)
Cloudera
Certified (QRC 1/2026)
Planned
Planned
Databricks
Certified (QRC 1/2026)
Certified (QRC 2/2026)
Certified (QRC 2/2026)
Snowflake
Certified (QRC 2/2026)
Certified (QRC 2/2026)
Certified (QRC 2/2026)
Authentication is currently limited to bearer-token (OAuth). Full credential flows (e.g., automatic token acquisition and refresh by negotiating directly with the identity provider) are not implemented inside SAP HANA Cloud. Instead, the credential flow is expected to live in the application layer, while the database layer simply consumes the access token.
For renewing access tokens, the SET SESSION CREDENTIAL statement can be used to inject a freshly obtained access token from the application layer:
-- renew the access token from the application layer
SET SESSION CREDENTIAL FOR REMOTE SOURCE 
  TYPE 'OAUTH' USING 'access_token=';
In other words, our expectation is that the credential flows are implemented in the application layer, while the database layer supports the way of renewing access tokens.
Please refer to the links below for further details.
 
More Improvements Worth Highlighting
Beyond the two main pillars above, several smaller but meaningful improvements landed alongside this Iceberg journey. They are bundled here for completeness.
Snapshot replica toggling for Apache Iceberg (QRC 1/2026)
For customers who want better query performance on Iceberg tables, snapshot replica toggling is now supported when the source is an Apache Iceberg table. As with other SDA-based virtual tables, you can toggle a virtual table to a snapshot replica by adding a snapshot replica with the ALTER VIRTUAL TABLE statement, then refresh or drop it as needed:
-- toggle to a snapshot replica
ALTER VIRTUAL TABLE DEMO.VT_ORDERS ADD SHARED SNAPSHOT REPLICA;

-- refresh the snapshot replica with the latest data from the source
ALTER VIRTUAL TABLE DEMO.VT_ORDERS REFRESH SNAPSHOT REPLICA;

-- delete the replica
ALTER VIRTUAL TABLE DEMO.VT_ORDERS DROP REPLICA;
A couple of important constraints to keep in mind:
Snapshot replica toggling always creates a full snapshot of the data the virtual table is pointing to. This means it does not help if the target Iceberg table holds a vast volume of data that cannot be replicated in full into SAP HANA Cloud. In such cases, the recommended approach is to keep federation only, or wait for the upcoming chunk-based replication discussed in Looking Ahead.
Unlike Delta tables, real-time replication via toggling is not yet supported for Iceberg. Real-time replication on Iceberg will become available together with the upcoming interval-based CDC support (see Looking Ahead).
Data Page V2 support in the Parquet reader (QRC 1/2026)
The Parquet ecosystem has been moving toward Data Page V2 (page_type=DATA_PAGE_V2), and several modern engines, now produce Parquet files using this newer page format. Starting with QRC 1/2026, the SAP HANA Cloud Parquet reader supports DATA_PAGE_V2, ensuring smooth interoperability with these engines. This is a quiet but important enabler for the full Iceberg story, especially for the Apache Iceberg REST Catalog scenario with Snowflake-managed Iceberg.
AliCloud Object Storage Service (OSS) support (QRC 2/2026)
The list of supported external object storages keeps growing. With QRC 2/2026, AliCloud Object Storage Service (OSS) joins Amazon S3, Azure Blob/ADLS Gen2, and Google Cloud Storage as a first-class storage option for SQL on Files.
However, please note one limitation specific to AliCloud deployments:
SQL on Files queries accessing Google Cloud Storage on AliCloud Cloud deployments are not supported.
Please refer to the links below for further details.
 
FAQs
Q: Can I write to an Apache Iceberg table from SQL on Files?
A: No. SQL on Files is read-only by design, both via the file adapter and the icebergcatalog adapter. INSERT, UPDATE, and DELETE on virtual tables pointing to Iceberg are not supported. Write operations should be performed by Iceberg-native engines (e.g., Spark, the catalog provider's compute) on the source side.
Q: Which Iceberg REST Catalog providers are officially certified?
A: Cloudera and Databricks running on AWS were initially certified with QRC 1/2026. With QRC 2/2026, Databricks is additionally certified on Azure and GCP, and Snowflake is newly certified across AWS, Azure, and GCP. Cloudera certification on Azure and GCP is planned for a future release. See the certification table in the icebergcatalog adapter section above for the full per-QRC, per-hyperscaler matrix. Other spec-compliant REST catalogs may work but are not officially certified.
Q: What kind of authentication is supported for the icebergcatalog adapter?
A: Bearer-token (OAuth) only. Full credential flows are expected to be handled in the application layer, with the access token being injected into SAP HANA Cloud either at remote source creation time or via SET SESSION CREDENTIAL for refresh.
Q: Can I replicate an Apache Iceberg table into a SAP HANA table?
A: Today, only snapshot replica toggling is supported for Iceberg. Full remote subscription-based replication (initial load + interval-based CDC), equivalent to the Delta table replication introduced in QRC 4/2025, is on the roadmap (see Looking Ahead).
Q: I'm running Google Cloud Storage on AliCloud. Can I use SQL on Files to read data from there?
A: No. SQL on Files queries accessing Google Cloud Storage are not supported on AliCloud deployments. Other supported object storages remain accessible.
Q: Is there any plan to support Apache Iceberg format V3?
A: The Iceberg ecosystem is evolving toward V3, and we are actively defining our adoption strategy. See Looking Ahead for the high-level direction.
Q: I'm already accessing Apache Iceberg tables via the file adapter. Do I need to change anything with the introduction of the icebergcatalog adapter?
A: No. The two adapters cover different scenarios and coexist. If you have been using the file adapter to access Iceberg tables directly on object storage, your existing setup keeps working as-is. The icebergcatalog adapter is an additional option for customers who manage their Iceberg tables through an Iceberg REST Catalog (Cloudera, Databricks, or Snowflake).
 
Looking Ahead
While this blog focuses on what is generally available with QRC 2/2026, there are a couple of directions we are actively exploring for the longer term:
Apache Iceberg table-to-HANA replication: Today, Delta table replication (QRC 4/2025) gives customers chunk-based or one-step initial load combined with user-driven or scheduled CDC. We see strong demand for the equivalent capability on Apache Iceberg via both the file adapter and the icebergcatalog adapter, including Data Replication UI integration. Release timing has not yet been finalized.
Apache Iceberg format V3: V3 introduces extended types (e.g., variant, geometry, geography, nanosecond timestamps), default values, row lineage, and binary deletion vectors. We plan to selectively enable V3 capabilities as the broader ecosystem matures, so customers get predictable behavior with V3 tables. Release timing has not yet been finalized.
These items reflect long-term direction rather than committed deliveries, so please treat them as forward-looking signals rather than firm release plans.
 
Conclusion
With SAP HANA Cloud QRC 2/2026, the Apache Iceberg story in SQL on Files is now complete in its first chapter: direct read-only access to Iceberg tables via the file adapter, enterprise-grade integration with Iceberg REST Catalogs via the new icebergcatalog adapter across AWS, Azure, and GCP, plus a set of supporting improvements like snapshot replica toggling, Data Page V2, and AliCloud OSS coverage.
This continues SAP HANA Cloud's commitment to open table formats, meeting customers where their data already lives, whether that's a Delta Lake, an Iceberg lakehouse, or a mixture of both. As we remain committed to innovation, stay tuned for upcoming updates that will continue to expand and enrich your SAP HANA Cloud experience, including richer Iceberg replication and evolving Iceberg V3 support.