Crunchy Data Warehouse Postgres
published on 2024/11/21
- Iceberg tables in PostgreSQL: You can create, manage, query, and update Iceberg tables that are cheaply and durably stored in S3 as easily as PostgreSQL tables, with fast analytical queries, and… perform ACID transactions that span across your operational tables and data lake 🤯.
- High performance analytics: Crunchy Data Warehouse extends the PostgreSQL query planner to delegate part of the query to DuckDB for vectorized execution, and automatically caches files on local NVMe drives. Together these optimizations deliver on average over 10x better performance than PostgreSQL (tuned) in TPC-H queries on the same machine, and even greater improvements on many common query patterns.
- Query raw data files in your data lake: Most data lakes consist of CSV/JSON/Parquet files in S3, which are passed between different systems. You can easily query data files and directories that are already in S3, or insert them into Iceberg tables. You can also query external Iceberg tables, Delta tables, and various geospatial file formats.
- Flexible data import/export: You can load data directly from an S3 bucket or http(s) URL into Iceberg or regular PostgreSQL tables, and you can write query results back to S3 to create advanced data pipelines.
- Seamless integrations: Crunchy Data Warehouse follows the “Lakehouse” architecture and brings together the Iceberg and PostgreSQL ecosystems. External tools can interact with Iceberg tables via PostgreSQL queries, or retrieve data directly from storage.