ClickHouse Observability Prices Logs and Traces at $0.20 per GB

ClickHouse Observability is a managed OpenTelemetry backend in public beta that stores logs, traces and metrics.

At standard rates, ClickHouse charges $0.20 per uncompressed gigabyte (GB) of logs and traces ingested. Metrics cost $0.10 per million data points. Both rates are 50 percent lower during public beta.

The service is powered by ClickHouse Cloud and open-source ClickStack, and you keep your existing OpenTelemetry collectors. It is built for site reliability engineers (SRE) and their artificial intelligence (AI) agents, as of 10 November 2026.

What ClickHouse Observability Is

ClickHouse Observability provides a fully managed telemetry storage backend built on ClickHouse Cloud and OpenTelemetry.

According to the launch post published by Mike Shi on 9 November 2026, the service is aimed at SREs and their software agents. You point your existing OpenTelemetry collectors toward a managed OpenTelemetry Protocol (OTLP) endpoint. Once data flows, ClickHouse handles data ingestion, schemas, and storage.

ClickHouse states that you do not need to size compute clusters, tune database engines, or manage ingestion pipelines. The service packages these backend tasks behind volume pricing instead of compute and storage.

ClickHouse lists two standard unit rates for the managed service: $0.20 per uncompressed GB for ingested logs and traces, and $0.10 per million data points for ingested metrics. During the public beta period, ClickHouse discounts both unit rates by 50 percent, bringing logs and traces to $0.10 per GB and metrics to $0.05 per million data points.

ClickHouse includes 30 days of retention during the public beta. It plans longer retention for general availability (GA) and has published no GA date. For teams deciding what to monitor, see our page on LLM usage monitoring.

How Pricing Works for GB Ingested and Data Points

ClickHouse prices the observability platform purely by the volume of telemetry data ingested into the system.

Logs and traces are billed on the uncompressed volume ingested each month, and metrics on the data points ingested each month. ClickHouse says charging metrics by data points rather than unique time series means adding useful dimensions does not introduce a separate cardinality charge.

ClickHouse includes 30 days of retention during the public beta and plans longer retention for general availability. It has not published a longer option for the beta, a general availability date, or the general availability retention period. ClickHouse also says on its product page that there is no monthly minimum.

The launch post names logs and traces in the $0.20 per GB row. The product page lists logs, traces, and session replays in that same $0.20 per GB row.

On our reading of ClickHouse's pages as of 10 November 2026, cost scales with the volume you ingest. For example, 100 GB of logs and traces in a month costs 100 times $0.20. That is $20 at the standard price and $10 at the public-beta price. 100 million metric data points cost 100 times $0.10. That is $10 at the standard price and $5 during public beta. This is arithmetic on ClickHouse's published prices, not an estimate of typical team usage.

ClickHouse says running observability on ClickHouse Cloud used to mean paying for compute and storage. It says that gave teams control but made the bill depend on understanding the resources a workload needs. The new service charges by volume instead.

How Telemetry Gets In Through OpenTelemetry and the ClickStack UI

You send telemetry directly to ClickHouse Observability by routing data through standard OpenTelemetry collectors to a managed OTLP network endpoint.

Engineering teams do not need to rewrite their instrumented applications or swap out deployed collector pipelines. You point existing OpenTelemetry collectors toward the managed endpoint, and ClickHouse handles ingestion, schemas, and storage.

Once telemetry arrives, teams investigate system behavior inside the ClickStack user interface. The UI lets you search logs, explore metrics, and follow traces. Operators can also turn useful queries into dashboards and alerts.

For organizations already operating an active ClickHouse cluster, ClickHouse documents a getting started workflow that lets administrators turn on observability capabilities directly within their existing service.

How Agents Read the Data Through MCP and SQL

The included ClickStack Model Context Protocol (MCP) server and Structured Query Language (SQL) are the two ways ClickHouse says agents read the data.

ClickHouse states that the ClickStack MCP integration lets AI agents search through logs, inspect service traces, identify recurring event patterns, and isolate behavioral deltas. By querying through SQL, agents and engineers can join signals and compare historical behavior.

ClickHouse also announced Observability Apps as a private preview feature. This capability lets users describe a monitoring interface in plain language, after which an agent creates corresponding queries, customized visualizations, and linked dashboard views. ClickHouse specifies that any generated code executes within an isolated sandbox environment, and all triggered queries remain read-only while inheriting the active user's database permissions. Observability Apps remain limited to private preview and are not part of the public beta tier.

ClickHouse describes the workflow in three steps. Send OpenTelemetry data to a managed OTLP endpoint. ClickHouse handles the infrastructure. You inspect logs, metrics, and traces in the ClickStack UI, through MCP tools, and with SQL.

What Happens to Managed ClickStack and Open-Source ClickStack

Managed ClickStack becomes a self-configured deployment option within ClickHouse Observability, while open-source ClickStack remains freely available for self-hosted installations.

The self-configured option gives teams control over database layouts, schemas, tuning, and compute, while ClickHouse Cloud manages the infrastructure. ClickHouse says it suits teams with database expertise and enough scale to justify the engineering work, in pursuit of the lowest unit costs.

ClickHouse says open-source ClickStack stays open source and that it remains committed to its development. You can deploy ClickStack on your own infrastructure and connect agents through MCP and SQL.

Who Should Skip ClickHouse Observability

ClickHouse Observability is the wrong choice for teams that require data retention past 30 days before the general availability release arrives.

ClickHouse includes 30 days of retention during the beta and has not published a longer option for it. The service also does not suit teams that want to control schemas, database tuning, and compute. Those teams are better served by the self-configured path or self-hosted ClickStack.

For what to send to a log store, see our AI agent audit log checklist.

Our verdict would change if ClickHouse releases general availability terms with longer retention or different pricing.

Our Takeaway on ClickHouse Observability

On our reading of ClickHouse's pages as of 10 November 2026, ClickHouse Observability fits teams already invested in OpenTelemetry that want managed logs, traces, and metrics. It also gives agents access through MCP and SQL.

ClickHouse prices it at $0.20 per GB of logs and traces and $0.10 per million metric data points, currently half that during public beta. Beta retention is 30 days.

It does not fit teams that need long retention today or want to control schemas and compute. General availability pricing and retention will decide whether it suits them.

To evaluate the fit for your environment, calculate your monthly ingestion volume in uncompressed gigabytes and review the projected costs against ClickHouse Observability pricing before launching a migration.