Google AlloyDB can query Iceberg tables with live data
Google says that on a 1 TB dataset its direct Iceberg access is as much as 7x faster than the comparable Amazon service.
Google Cloud announced direct Iceberg access in AlloyDB for PostgreSQL, in preview. AlloyDB is the managed database of Google. Apache Iceberg is an open table format for the history data in a lakehouse, a store for analytics data. Developers can connect Iceberg tables with live data in one SQL query. Google also made AlloyDB sync and Lakehouse federation available to all users.
Three ways in
- Direct Iceberg access: read tables where they are.
- AlloyDB sync: copy BigQuery data into AlloyDB on a schedule you set.
- Lakehouse federation: query BigQuery tables where they are, and BigQuery does the work.
The numbers
Google says that in its test, direct access is as much as 7x faster than the comparable PostgreSQL service of Amazon. It is as much as 3.5x faster than the open source PostgreSQL extension. Google gave no test method. Google says Etsy made pipeline queries 60% faster on three petabytes.
What is not known
Direct access is in preview. A synced copy is only as new as its last refresh. The source gives no prices.
Insight: Two ways to remove the data pipeline for agents
Google keeps the history in the lakehouse and reads it from the database. MariaDB, in a launch this month, puts agent memory, a cache, live analytics and vector search in one system. The two sell agents with no custom pipeline, but Google bets that data stays where it is and MariaDB bets on one system. Select by where your data is today.
Sources
Posted
