ClickHouse has announced three additions to its Postgres offering: an open-source replication engine called WalShadow, a private preview of its managed Postgres service on Google Cloud, and an extension for importing and exporting cloud-storage data. The most striking claim is that WalShadow can make Postgres changes visible in ClickHouse in around 200 milliseconds, rather than the roughly 10 seconds ClickHouse reports for its PeerDB comparison.
ClickHouse Watch analysis
What happened
In a post published on 13 October, ClickHouse says WalShadow reads Postgres’s physical write-ahead log, or WAL, and replicates changes into ClickHouse. The company says it is open source, supports Postgres 17 and 18, and is available for testing through a Docker quick start. Its managed version is in private preview for ClickHouse Managed Postgres customers; the company says the feature requires physical WAL access that most managed Postgres services do not expose.
ClickHouse’s benchmark on modest eight-vCPU instances reports around 200 milliseconds from a Postgres commit to visibility in ClickHouse, against around 10 seconds for PeerDB. It also reports throughput of about 300,000 rows per second, compared with PeerDB’s 120,000. These are company benchmarks, and ClickHouse provides a benchmark harness for others to reproduce them.
The same announcement puts ClickHouse Managed Postgres into private preview on Google Cloud, with deployment in regions where ClickHouse Cloud runs. A separate chdb Postgres extension brings imports and exports from services including Amazon S3, Google Cloud Storage and Azure Blob Storage into Postgres’s COPY command. ClickHouse says its NYC Taxi dataset import test was more consistent than the other extensions it tested; two DuckDB-based extensions took around two to three times as long on CSV, JSON and Parquet files.
Why it matters
These additions tackle the unglamorous joins between transactional databases, analytical queries and cloud files. Faster replication could make recent Postgres data available for analysis sooner, while the Google Cloud preview gives teams another place to run both sides of the stack. The import extension aims to make moving data less of a bespoke pipeline project.
The headline speed figures are worth testing, not treating as universal results. They come from ClickHouse’s own comparisons, and WalShadow’s managed preview is limited to ClickHouse Managed Postgres. Still, open-source code and a benchmark harness give developers a way to inspect the approach rather than merely admire the numbers.
Our read
This is a substantial product update because it gives database teams concrete new tools, not just a fresh diagram of a “unified” stack. WalShadow is the standout: if independent tests confirm the latency and throughput claims, it could make near-real-time analytics more practical for workloads already using Postgres. Teams should check the code and benchmark their own workloads before planning around the results.
What to watch
- Whether independent tests reproduce WalShadow’s latency and throughput figures.
- When WalShadow and ClickHouse Managed Postgres on Google Cloud move beyond private preview.
- How the chdb extension performs on datasets and storage services beyond ClickHouse’s tests.
Discussion spark: Would you trust ClickHouse’s WalShadow benchmark as a reason to try it now, or wait for independent results before changing a production data pipeline?
Sources and evidence
- Source update (13 October 2026, 15:35 UTC)
not affiliated with or endorsed by ClickHouse