IT Brief New Zealand - Technology news for CIOs & IT decision-makers
New Zealand
Google Cloud adds Iceberg access to AlloyDB for PostgreSQL

Google Cloud adds Iceberg access to AlloyDB for PostgreSQL

Wed, 18th Nov 2026 (Today)
Mara Sugue
MARA SUGUE News Editor

Google Cloud has introduced direct Apache Iceberg access in AlloyDB for PostgreSQL and made AlloyDB sync and Lakehouse federation generally available.

The new Iceberg feature, now in preview, lets developers query Apache Iceberg tables stored in cloud storage directly from AlloyDB and join them with operational data in a single SQL query.

With these changes, AlloyDB now offers three ways to work with data across operational databases and analytical systems: direct Iceberg access for tables stored in cloud storage, AlloyDB sync for copying selected lakehouse data into AlloyDB on a schedule, and Lakehouse federation for querying BigQuery tables in place.

The move reflects a broader shift in data architecture as companies try to combine live transactional records with larger stores of historical data without maintaining separate pipelines between systems. Google Cloud is positioning AlloyDB as a database that can sit alongside a lakehouse or data warehouse while still handling operational workloads.

Three options

With direct Iceberg access, users register an Iceberg table once and query it through PostgreSQL. They can then join that table with live operational data already held in AlloyDB.

AlloyDB reads Iceberg tables directly from Cloud Storage using either a bucket path or the Lakehouse runtime catalog. Queries run on the same database instance as the application, rather than requiring a separate reader instance.

In internal testing on a 1 TB dataset in Cloud Storage, direct Iceberg access was up to seven times faster on aggregation queries than Amazon's comparable PostgreSQL-compatible service and up to 3.5 times faster than an open-source PostgreSQL extension, according to Google Cloud.

AlloyDB sync addresses a different use case. Instead of querying data where it sits, it copies selected data from BigQuery into AlloyDB either once or on a schedule.

That approach is intended for data that applications access frequently, where lower-latency local reads matter more than querying a remote analytical store in place. The synced copy becomes a local AlloyDB table that users can index and serve through read pools.

Lakehouse federation, also now generally available, lets AlloyDB query native BigQuery tables directly and combine the results with operational records held in AlloyDB. In this model, BigQuery carries out most of the processing and returns only the results to AlloyDB.

Users can also choose which project is billed for those queries and restrict which roles can use them.

Operational use

Google Cloud outlined several examples of combining operational and historical data. These included fraud checks that compare a live payment with years of past customer transactions, retail personalisation based on a shopper's current basket and lifetime value, and pricing decisions that join live inventory with longer-term demand or weather history.

Another example involved data teams publishing curated tables once to Iceberg and letting multiple applications read them directly, rather than maintaining separate copies for each system.

This model allows applications and software agents to act on operational and lakehouse data together while relying on the permissions already granted in AlloyDB, Google Cloud said.

Customer reference

Google Cloud pointed to Etsy as one example of a business using a borderless lakehouse approach. According to the company, Etsy unified three petabytes of data and improved pipeline query speeds by 60% by joining data in place.

It also cited Loyal Guru as an existing user of AlloyDB and BigQuery together through this type of arrangement.

“AlloyDB, along with other Google Cloud products like BigQuery, provides the agility and performance needed to continually enhance our platform's capabilities and help us anticipate emerging trends rather than merely reacting,” said Javi Fernández, Co-founder and CTO, Loyal Guru.