ClickHouse Managed Postgres Is Generally Available on AWS

ClickHouse Managed Postgres is a Postgres service that runs on local nonvolatile memory express (NVMe) storage. It reached general availability (GA) on Amazon Web Services (AWS) on 9 November 2026.

ClickHouse documents branches restored from point-in-time recovery in several minutes to tens of minutes. Instant copy-on-write branching from Xata is planned for "the coming months", so teams that need instant branches today use Xata directly. The instant clones and the artificial intelligence (AI) agent tier are not active in ClickHouse Managed Postgres today.

What ClickHouse Managed Postgres Is

ClickHouse Managed Postgres is an operational database service running on local NVMe storage that reached general availability on AWS on 9 November 2026.

On Google Cloud (GCP), ClickHouse Managed Postgres remains in private preview, with ClickHouse stating that general availability on GCP is coming soon. The underlying service is built with Ubicloud.

Alongside the general availability announcement, ClickHouse announced its acquisition of Xata. Xata develops database branching, copy-on-write storage, scale to zero, and automatic idling for Postgres instances.

ClickHouse states that it will fully integrate Xata into ClickHouse Managed Postgres in the coming months. That work targets two specific paths: instant branches and clones for production NVMe-backed Postgres, and a dedicated tier for AI agents focused on sandboxing, memory, and agent-first workloads.

Instant branching is not in ClickHouse Managed Postgres today, so a team that needs instant branches immediately must use Xata directly. ClickHouse documents point-in-time-recovery branches that restore from backups in several minutes to tens of minutes. ClickHouse says it is actively onboarding customers to Xata while the integration is built.

What Is Generally Available Today

High availability, connection pooling, and replication tools are generally available on AWS today.

ClickHouse reports that the service now supports hundreds of production customers across new applications and migrations from other managed Postgres providers. The workloads span from AI-native startups to enterprises operating multi-terabyte databases. ClickHouse claims that local NVMe storage delivers 2 to 10X performance improvements for online transaction processing (OLTP) workloads. It also claims up to 10X improvements and better reliability over the past 10 months. Those figures are ClickHouse's own claims, not independent benchmark measurements.

High availability configurations provide up to two standby instances across availability zones with automatic failover. Point-in-time recovery and automated backups protect operational data from accidental deletions or corruption. Integrated PgBouncer connection pooling handles large volumes of client connections, while dedicated read replicas offload read-heavy traffic from primary database nodes.

Operational monitoring includes Query Insights, query logs, system metrics, and an exportable Prometheus endpoint. Security controls include Security Assertion Markup Language (SAML) single sign-on (SSO), role-based access control (RBAC), internet protocol (IP) address allow-listing, private network connectivity, and encryption at rest and in transit. Teams can automate cluster management through OpenAPI, Terraform, or the ClickHouse command-line interface (CLI).

To connect transactional data with analytical systems, ClickHouse provides two data movement paths. WalShadow reads the physical write-ahead log (WAL) for sub-second replication, and ClickHouse says commits reached ClickHouse in around 200 milliseconds in its benchmarks. WalShadow is currently in private preview. For general use, ClickPipes provides a change data capture (CDC) alternative at no additional cost. Replication from Managed Postgres to ClickHouse is free with either mechanism. Applications can also use the pg_clickhouse extension to query ClickHouse through standard Postgres SQL interfaces.

The core capabilities available at general availability on AWS include:

• Deploy up to two standby nodes across availability zones with automated failover.
• Restore databases to a specific timestamp with automated backups and point-in-time recovery, for example after an accidental deletion.
• Pool client connections using integrated PgBouncer instances without running separate proxy servers.
• Route analytical and read-heavy queries across dedicated read replicas.
• Replicate tables into ClickHouse Cloud without fees using ClickPipes CDC or the private preview WalShadow engine.
• Federate queries between Postgres and ClickHouse using the pg_clickhouse extension.
• Enforce access controls using SAML SSO, RBAC policies, IP allow-lists, and encryption at rest and in transit.
• Manage environments programmatically using OpenAPI endpoints, Terraform scripts, or the ClickHouse CLI.

What Changed on Price at General Availability

With the arrival of general availability, ClickHouse ended its public beta discount, and charges for backup storage and public network egress became billable.

ClickHouse stated that it expects these additional operational charges to be modest for most customers. New ClickHouse Cloud registrations receive $300 in free trial credits valid for 30 days.

On its official ClickHouse pricing page, ClickHouse publishes Managed Postgres compute and storage starting from $0.056 per unit per hour. The pricing page states that the service is available on all three platform tiers, accompanied by an interactive monthly cost estimator calculator. Neither the pricing page nor the documentation defines what a unit represents, so users should not attempt to calculate raw hourly server costs without using the calculator.

According to the official Managed Postgres pricing documentation, pricing is calculated on the full virtual machine (VM) configuration rather than separate compute and disk billing line items. The cost includes the central processing unit (CPU), random-access memory (RAM), and NVMe storage allocated to the virtual machine.

ClickHouse says over 50 configurations are available. The smallest has 1 virtual CPU (vCPU), 8 gigabytes (GB) of RAM, and 59 GB of NVMe storage. The largest has 96 vCPUs, 768 GB of RAM, and 60 terabytes (TB) of NVMe storage. Pricing, limits, and features also vary across the Basic, Scale, and Enterprise tiers. For exact figures, ClickHouse directs users to its pricing calculator, which takes the tier, configuration, and standby setup.

Xata Branching and the ClickHouse Roadmap

Branching creates a copy of a database in place of a slow restore. Xata pairs it with automatic idling and scale to zero.

Xata claims that creating a database branch takes only around 3 seconds, even on databases with multiple terabytes of data. Xata says this works because copy-on-write branching does not copy the data, and the child branch has its own compute on a different physical node.

Xata also claims that inactive database branches automatically scale to zero and idle, waking up in approximately 1 second when new queries arrive on Xata Cloud. Xata does not bill per user or per branch, and users pay only for storage differences written to the branch. On the Xata pricing page, Xata Cloud starts at $0.012 per hour plus $0.28 per GB per month for storage. The open-source self-hosted version of Xata is free, with wake-up times taking roughly 10 seconds compared to 1 second on Xata Cloud.

Xata stated that nothing changes for existing Xata customers in the short term, and new accounts continue to be accepted. Xata confirmed it will eventually migrate its customers over to ClickHouse Managed Postgres once the integration is ready.

ClickHouse plans two paths for Xata technology. One is instant copy-on-write branches and clones for local NVMe-backed Postgres. The other is a dedicated agent tier for agent sandboxing, agent memory, and agent-first workloads. ClickHouse has not published a release date beyond "in the coming months", and it has not published pricing for the agent tier.

On our reading of the ClickHouse and Xata announcements as of 10 November 2026, available and planned features split as follows.

Who Should Skip ClickHouse Managed Postgres

Teams that need instant database branching this quarter should use Xata directly rather than wait for ClickHouse Managed Postgres.

Teams on cloud providers other than AWS should also pause. ClickHouse lists AWS as generally available and GCP as private preview. Teams that need a fixed monthly quote before trying the service should look closely at the billing. ClickHouse publishes a starting figure of $0.056 per unit per hour and relies on its pricing calculator across more than 50 configurations and three tiers. Exact monthly pricing needs the vCPU, RAM, storage, and standby setup you choose.

For teams choosing a store for search workloads, see our comparison of the best vector databases. ClickHouse's extensions list includes pgvector for vector search.

The recommendation would change if ClickHouse ships the native Xata branching integration or the agent tier, publishes agent-tier pricing, or changes the starting price.

Our Takeaway on ClickHouse Managed Postgres

On our reading of the ClickHouse and Xata announcements as of 10 November 2026, ClickHouse Managed Postgres is a Postgres service on local NVMe storage. ClickHouse says it is generally available on AWS.

The evidence basis rests on reported announcements, documentation, and pricing sheets published by ClickHouse and Xata, alongside our labelled synthesis of their stated roadmaps. Performance claims of 2 to 10X improvements remain vendor claims that have not been independently benchmarked.

The main exception is any team that needs instant branching, scale to zero, or an agent tier today. Those features are at Xata and are planned for ClickHouse Managed Postgres. ClickHouse's own branches restore from point-in-time recovery in minutes. Teams that need instant branching today use Xata directly.

Before committing your infrastructure, run ClickHouse's pricing calculator for your exact vCPU and storage configuration to evaluate the total operational cost of ClickHouse Managed Postgres.