Building a Modern Data Lakehouse on AWS: S3, Iceberg, Glue, Athena, and Lake Formation

In 2026, the data lakehouse architecture has become the default choice for analytics on AWS. This hybrid solution merges the cost-effective, schema-flexible storage of a data lake (Amazon S3) with the performance, ACID transactions, and governance features of a data warehouse — all without the need to run a traditional warehouse for most workloads.

The key enabling technology behind this architecture is Apache Iceberg, an open table format that provides SQL-like capabilities (INSERT, UPDATE, DELETE, time travel) to files stored in S3.

Key components of the AWS data lakehouse architecture include:

1. Consumption Layer: Query engines such as Athena, Redshift Spectrum, EMR (Spark), and QuickSight (Business Intelligence) can access and analyze data directly from S3.

2. Governance Layer: AWS Lake Formation provides fine-grained access control, audit trails, and overall governance of the data lakehouse.

3. Table Format: Apache Iceberg serves as the table format that brings ACID transactions, time travel, schema evolution, and other warehouse capabilities to files stored in S3.

4. Processing Layer: ETL (Extract, Transform, Load) tasks are handled by AWS Glue, which uses Spark for processing. Glue Streaming supports real-time data processing, while EMR (Elastic MapReduce) provides a flexible cluster computing service. Zero-ETL and Firehose are additional options for data ingestion without manual configuration.

5. Ingestion Layer: Data ingests into the data lakehouse through various methods, including Kinesis (real-time data streams), Data Migration Service (DMS) for data migration, Application Integration (AppFlow) for cloud-based applications, S3 Transfer (direct file transfer), and Direct PUT for manual file uploads.

6. Storage Layer: Amazon S3 serves as the primary storage layer, accommodating raw, curated, and analytics zones for data organization. Each zone serves a specific purpose, such as landing data in its original format or storing cleaned, validated data in Iceberg format.

Iceberg's role in the AWS data lakehouse architecture is crucial, as it enables SQL-like capabilities on top of S3 files. By providing row-level operations, ACID transactions, schema evolution, time travel, and automated partition handling, Iceberg enhances the flexibility and performance of data storage and querying in the data lakehouse environment.

Creating an Iceberg table in Athena involves defining a table schema, specifying partitioning, and setting appropriate table properties. For example:

CREATE TABLE analytics.orders (

order_id STRING,

customer_id STRING,

amount DECIMAL(10, 2),

status STRING,

order_date TIMESTAMP,

region STRING

) PARTITIONED BY (region, month(order_date))

LOCATION 's3://my-lakehouse/analytics/orders/'

TBLPROPERTIES (

table_type = 'ICEBERG'

);

In summary, the AWS data lakehouse represents a modern, efficient, and cost-effective approach to big data analytics. By leveraging the combined strengths of S3, Iceberg, Glue, Athena, and Lake Formation, organizations can build a robust analytics platform that maximizes flexibility, performance, and governance without the overhead of managing a traditional data warehouse.