Snowflake Postgres Unifies Your Apps, Analytics and AI

Every enterprise runs in two worlds. In one, your apps process orders, track events and serve customers in real time. In the other, your analytics platform uncovers insights, trains models and powers AI. Between them lies a tangled mess of extract, transform, load (ETL) pipelines, batch jobs and third-party tools that cost a fortune to maintain. 

Now, Snowflake Postgres is closing that gap. At Snowflake Summit, we announced:

• Data mirroring: Always-on replication between Postgres and Snowflake (public preview soon) 

• Postgres for your data lake: A more flexible way to sync Postgres with analytics using open formats like Iceberg (generally available soon) 

They provide a seamless connection between transactional and analytical data, no complex pipelines required.

Tackling the No. 1 infrastructure problem

Customers consistently tell us that moving data between online transaction processing (OLTP) and online analytical processing (OLAP) databases is the most painful infrastructure task in their data estate. The visible costs, like ETL licensing, pipeline compute and connector fees, are just the tip of the iceberg. Underneath lurk data inconsistencies, governance risks and engineering hours burned on maintenance and the delayed decisions that come from stale data.

In the era of AI agents and real-time apps, this approach leaves you always a step behind. Your fraud model can't catch today's threat with last night's batch load. Your pricing engine can't optimize with a six-hour lag.

With Snowflake Postgres, we’ve developed a fundamentally different and radically simpler approach to using Postgres and Snowflake together.

Two new ways to connect your data: Always-on data mirroring and open-format data lake integration

Powered by our open source pg_lake extension, you can now choose between always-on data mirroring and flexible, open-format data lake integration to connect transactional and analytical data seamlessly.

1. Data mirroring: ‘Set it and forget it’ data replication

Data mirroring provides low-latency replication between Postgres and Snowflake. Once you create a mirror, Snowflake maintains target tables that reflect the current state of their source tables, including schema changes and new tables you create in mirrored schemas.

Set it up in a few clicks via Snowflake CoCo, the Snowsight UI or a single SQL command. That's it. Just your data, flowing where it needs to go. Check it out in this demo:

Key benefits of data mirroring include:

• Zero infrastructure to manage: Mirrors run entirely inside Snowflake. There's no external CDC service to deploy, no connector process to babysit and no additional vendor to manage.

• Always-fresh reads: Every mirrored table includes a $live view that combines already-applied data with in-flight changes. So readers see every committed source change within seconds.

• Transactional consistency: Changes from a single source transaction appear on the target together. Cross-table relationships (like foreign keys) stay intact, so your downstream joins and reports remain accurate.

• Built-in change history: Every mirrored table automatically gets a seven-day change feed ($changes) that exposes inserts, updates and deletes, queryable from both Snowflake and Postgres. 

• High throughput at any scale: Replication uses an optimized apply strategy that skips expensive full-table scans, so performance stays fast as your data grows.

Data mirroring is especially useful for teams who want to stop thinking about data movement. You set it up once, and your OLTP and OLAP stay in sync automatically, continuously and reliably. And data mirroring will soon work in the other direction, too: Snowflake-to-Postgres mirroring lands later this year, creating a true bidirectional bridge between your transactional and analytical worlds.

2. Postgres for your data lake: The data movement Swiss Army knife

Not every use case calls for continuous sync. Sometimes you need more flexibility. For example, you might want to move specific files, create shared open-format tables or transform data on its way to Snowflake.

Postgres for your data lake gives you that flexibility:

• File movement: Push and pull files between Postgres and Snowflake through internal Snowflake stages or external object storage.

• Shared Apache Iceberg™ tables: Create open-format Iceberg tables that both your Postgres and Snowflake can read. This means one table for two systems, and zero duplication.

• Transform as you move: Apply SQL transformations — filters, joins, aggregations — as data moves between Postgres and Snowflake. 

This gives developers the full Postgres experience they love, plus native interoperability with open standards like Iceberg and Parquet. It’s for teams that want more control over how, when and which data moves.

The result: One platform, zero pipelines

Together, these capabilities mean you can choose the right approach for each of your workloads.

Real results: Enterprises are already eliminating pipelines

Ericsson: From 48-hour lag to minutes

Ericsson's foundation data team collects software, hardware and license data from every customer network in the world. Multiple downstream teams — from AI to customer support — depend on this data. Previously, data was trapped in four legacy databases connected by pipelines that took up to 40 days to sync.

By consolidating onto Snowflake Postgres, Ericsson eliminated those pipelines entirely. Their customer support platform, where ticket creation depends on knowing exactly what's deployed in the field, went from 48-hour data lag and 12-hour processing times to under an hour, end to end. Now, every team accesses the same trusted data via Snowflake's sharing layer.

SimCorp made data synchronization 10x faster

SimCorp delivers investment software to the world’s leading financial institutions, managing extensive financial instrument data spanning more than 40 years. To support this scale, they operated separate Postgres and Snowflake environments, with custom reconciliation processes requiring hours of data transfer between systems.

With Snowflake Postgres, that complexity was eliminated. Snowflake Postgres removes the need for reconciliation code, streamlining operations and reducing overhead.

The impact has been significant. Their market ID service, which standardizes exchange codes to ensure consistent tracking and reporting, reduced synchronization time by 10x, from several hours to under 20 minutes. At the same time, their mission-critical risk models and credit spread curves, which enable investment teams to quantify market risk, now achieve immediate availability for global clients through AI and data sharing.

Looking ahead, SimCorp is building on this foundation to enable intelligent load balancing between Postgres and Snowflake through its data platform. This evolution is paving the way for a new generation of applications powered by integrated transactional and analytical data — driving faster insights, improved scalability and stronger outcomes for clients worldwide.

Why Snowflake Postgres is different

Most data replication tools work by wedging a separate service between your database and your warehouse, giving you another system to monitor and another vendor in your stack. They treat data movement as an external problem bolted onto your infrastructure.

Snowflake Postgres treats it as a native capability. Data movement is built on pg_lake, our open source Postgres extension that lets Postgres write directly to object storage — the same layer Snowflake reads from. There's no intermediary. Just Postgres and Snowflake, working together natively.

For you, this means:

• No new vendors to evaluate, pay for and manage

• No new infrastructure to deploy and monitor

• No new skills to learn — it's standard Postgres and standard Snowflake, built on open standards

• No data lag undermining your AI agents and real-time apps

And if you're already a Snowflake customer, you can power these transactional workloads from your existing committed spend, no new contract required.

When transactional and analytical data are one, AI and apps operate on the latest truth. Agents see the latest transactions in near real time. Pricing models run on fresh data. Fraud detection catches threats as they happen. Dashboards reflect current reality, not yesterday's reality.

Get started today

Data mirroring from Postgres to Snowflake is available in public preview soon. Postgres for your data lake is generally available soon. Check out these resources to get started:

• Getting Started with Snowflake Postgres

• Quickstart: Sync data with Iceberg and pg_lake

• pg_lake on GitHub

• Moving data with Snowflake Postgres

Forward-looking statements: This post contains forward-looking statements about future product offerings that are not commitments to deliver. Actual results may differ. See our latest 10-Q for more information.