pgEdge Launches ColdFront to Slash Postgres Storage Costs

ColdFront functions by automatically migrating older data partitions into Apache Iceberg format on S3-compatible storage. Because the system integrates the DuckDB vectorized engine directly into the PostgreSQL process, it enables analytical queries on cold data at speeds significantly faster than traditional row-based storage. The primary advantage for database administrators is the ability to maintain a single table name for both hot and cold tiers, requiring zero changes to existing application code or SQL queries.

The solution is designed to handle the full data lifecycle, including compliance-driven tasks like GDPR deletion requests, which can now be executed as standard SQL statements rather than requiring manual data rehydration. According to pgEdge, organizations can realize storage savings of up to 90% by moving infrequently accessed data off SSDs. The architecture remains vendor-neutral, utilizing stock PostgreSQL versions 16 through 18 and ensuring that data stored in Iceberg remains accessible to external tools like Spark or Trino without dependency on pgEdge software.

Beyond storage efficiency, the tool supports a decoupled mode where PostgreSQL acts as a stateless compute layer for AI and machine learning pipelines. This allows teams to query massive historical datasets for RAG retrieval or agentic analytics without building secondary data warehouses. ColdFront is available now as a production-grade beta and will be integrated into pgEdge’s enterprise and cloud offerings later in 2026.