--- title: ClickHouse Observability description: 'Fully managed, full-fidelity observability for AI agents and SREs: keep every log, metric, and trace with fast queries over high-cardinality data and low-cost, long-term retention.' --- # Full-fidelity observability for AI agents and SREs No sampling. No roll-ups. No infra. - [Start free trial](https://console.clickhouse.cloud/signUp?intent=o11y&loc=clickhouse-observability-hero) - [Contact sales](https://clickhouse.com/company/contact?loc=clickhouse-observability-hero) - [Prefer open source? Explore ClickStack](https://clickhouse.com/clickstack?loc=clickhouse-observability-hero) Keep every log, metric, and trace. Query high-cardinality data fast, with low-cost, long-term retention. Send your OpenTelemetry data to one managed endpoint, and we’ll handle ingestion, storage, and scaling. - Full-fidelity telemetry and long-term retention - ClickHouse performance with simple volume-based pricing - Ingestion, schemas, storage, and scaling managed for you - Your agents, models, and investigation workflows Trusted for observability by Netflix, Shopify, Visa, Cloudflare, Sony, Comcast, Anthropic, Character.ai, Sierra, eBay, Cisco, Zoox, DoorDash, GitLab, IBM, SolarWinds and Vimeo. > During this past Black Friday and Cyber Monday, a minute of downtime can cost our merchants $5.1 million. That's why observability is critical, we need to detect issues and prevent any downtime. We saw around a 16x improvement out of the box with ClickHouse once we had it deployed, and at peak performance over 30x. And for us, less compute is money. > > — [Shopify](https://clickhouse.com/blog/shopify-observability-at-global-scale) > ClickHouse played an instrumental role in helping us develop and ship Claude models. With ClickHouse, our database is green, queries are lightning-fast, and money is no longer on fire. It’s already delivered tremendous value in helping us build state-of-the-art language models. > > — Anthropic > The cost of slow observability goes beyond infrastructure; it directly impacts customer experience. Mercado Libre’s migration to ClickHouse Cloud delivered 75x faster queries, efficient high-cardinality filtering, and 86–89% data compression across massive datasets, including 9.55 trillion spans and 181 TiB of raw data. > > — [Mercado Libre](https://clickhouse.com/blog/mercado-libre-observability-on-clickhouse-cloud) > Observability data keeps growing, with massive spikes from the simulations and testing we run to ensure vehicle safety. We’ve seen ingestion rates reach over 1.5 million logs per second, and ClickHouse handled that volume with just a couple of nodes, which was incredibly impressive. Now, with ClickStack, it’s even easier to query that data and quickly find what we’re looking for. > > — [Zoox](https://clickhouse.com/videos/openhouse-clickstack-linkedin-zoox) ## Built for SREs and their agents Give engineers and agents the data and query performance to test hypotheses, follow the evidence, and reach root cause even as systems grow in size and complexity. ### Fast, concurrent queries Test more hypotheses in parallel and uncover root causes sooner with fast queries across high-cardinality telemetry, without API limits slowing engineers or agents down. ### Full telemetry and history Correlate unsampled, high-cardinality telemetry with historical patterns, giving engineers and agents the evidence to reach root cause. ### Lower costs at scale Keep ingestion, retention, and queries affordable so agents can follow the evidence without rationing queries or discarding context. ### SQL agents already know Let agents test complex hypotheses with familiar, expressive SQL, exploring beyond predefined queries and fixed API operations. ## Keep the detail. Keep the history. Lower the cost. Keep the detail your investigations need with ClickHouse cost efficiency and straightforward volume-based pricing. Pay by the GB for logs, traces, and session replays, and by the million data points for metrics. No monthly minimum. - **Logs, traces, and session replays:** $0.10 per uncompressed GB, 50% off during beta (GA rate $0.20) - **Metrics:** $0.05 per million data points, 50% off during beta (GA rate $0.10) 30 days of retention included. No monthly minimum. No separate metrics cardinality charge. ### More telemetry for the same budget What $1,000 a month buys for logs, at GA rates (50% off during beta): - ClickHouse: 5 TB - New Relic: 2.5 TB, 2x less (excludes user seat costs) - Grafana Cloud: 1.82 TB, 2.8x less - Splunk Cloud: 0.46 TB, 11x less (assumes a 100 GB/day license) - Datadog: 0.39 TB, 13x less (assumes an average log line of 1 KB) Log ingestion only; metrics, traces, and session replays are not included. Competitor figures use indicative list pricing sourced from public pricing pages. ClickHouse Observability at the GA rate of $0.20 per GB uncompressed, ingested. ### Get started with ClickHouse Observability for free We'll get you started on a 30 day trial and $300 credits to spend at your own pace. [Start free trial](https://console.clickhouse.cloud/signUp?intent=o11y&loc=clickhouse-observability-economics) ## Send your data. We run the platform. Send your OpenTelemetry data to one managed endpoint and start investigating. Ingestion, storage, optimization, and scaling are handled for you. 1. **Connect your telemetry:** Send data from your existing OpenTelemetry SDKs and collectors. 2. **We manage the platform:** Ingestion, storage, optimization, and scaling are handled for you. 3. **Start investigating:** Explore data with search, dashboards, AI notebooks, or your agents. ClickHouse Observability is powered by [ClickStack](https://clickhouse.com/clickstack?loc=clickhouse-observability-platform), the open source observability stack for ClickHouse. ## Investigate together. Keep engineers in control. Explore an incident in a shared AI notebook that keeps evidence, queries, and reasoning together. Follow different hypotheses and guide the investigation as new information emerges. - **Branch your investigation.** Explore multiple hypotheses without losing the evidence or context from each path. - **Inspect each step.** Review the queries, reasoning, and visualizations, and use ClickHouse SQL to explore further. - **Continue together.** Share the investigation so another engineer can pick it up without rebuilding the context. ## Bring your own observability agents Connect Claude Code, Codex, Cursor, or your own agents to ClickHouse Observability through MCP, combining your telemetry with your code, runbooks, and operational context. Turn findings into saved searches, dashboards, and alerts so your team can review the evidence and build on each investigation. [Connect your agent](https://clickhouse.com/docs/clickstack/mcp) ## Give your agents tools built for investigation The ClickHouse Observability MCP server provides purpose-built tools for common investigation steps, with access to ClickHouse SQL when a question needs deeper analysis. Each tool encodes a common investigation workflow and an optimized query, so agents spend fewer steps reconstructing how to ask and more on the evidence. Tools include search, trace waterfalls, time series, event patterns, event deltas, SQL, alerts, saved searches, data sources and dashboards. [In our evaluations](https://clickhouse.com/blog/benchmarking-the-clickstack-mcp-server-with-hdx-evals) against a generic ClickHouse MCP server, agents using these tools were: - 17% more accurate - 25% fewer tool calls - 2.7× more consistent ## Enterprise observability, fully managed Focus on your systems and investigations. We handle the infrastructure, backups, upgrades, and maintenance. ### Built for reliability Replication across multiple availability zones provides high availability, with automated backups and built-in disaster recovery protecting your telemetry. ### Control access Manage who can access your telemetry and how they connect, with user and network access controls backed by comprehensive activity logging. ### Enterprise-grade security Built on ClickHouse Cloud, with encryption in transit and at rest and SOC 2 Type II compliance. - [Start free trial](https://console.clickhouse.cloud/signUp?intent=o11y&loc=clickhouse-observability-enterprise) - [Contact sales](https://clickhouse.com/company/contact?loc=clickhouse-observability-enterprise) ### Want more control? Choose self-managed. For engineering teams with database expertise seeking the lowest possible unit costs, ClickHouse Observability offers a **self-managed** experience. You control the configuration. ClickHouse Cloud manages the underlying infrastructure. [Why choose self-managed?](https://clickhouse.com/observability#faq-self-managed) - **Tune your telemetry.** Optimize schemas, compression, and indexes for your data. - **Right-size compute.** Match resources to your workload and budget. - **Scale workloads independently.** Configure separate resources for ingestion, queries, and alerting. ## From the first signal to the root cause Find the error, follow the request, and understand what changed. Explore connected telemetry with the tools your team needs throughout an investigation. Watch how engineers and agents move from an initial signal to the evidence behind it. [Video](https://www.youtube.com/watch?v=n0vO5Dy6FBA) ## Instrument your applications Already using OpenTelemetry? Keep your compatible SDKs and collectors and send telemetry to your managed endpoint. Starting fresh? Instrument your applications with OpenTelemetry to bring logs, metrics, and traces together. [Explore instrumentation guides](https://clickhouse.com/docs/clickstack/ingesting-data/overview) Supported sources include Node.js, Go, Java, JavaScript, Next.js, Python, Ruby, Cloudflare, Kubernetes, OpenTelemetry, AWS and Fluentd. ## Try ClickHouse Observability Give your SREs and agents full-fidelity telemetry, long-term context, and fast queries at ClickHouse economics, with the platform fully managed. - [Start free trial](https://console.clickhouse.cloud/signUp?intent=o11y&loc=clickhouse-observability-footer) - [Contact sales](https://clickhouse.com/company/contact?loc=clickhouse-observability-footer) ## FAQs Explore our FAQs to learn more about ClickHouse Observability, and if you don't see what you need, we're always happy to chat. [Ask us anything](https://clickhouse.com/company/contact) ### What is ClickHouse Observability? ClickHouse Observability is a fully managed observability service for engineers and their agents. Investigate logs, metrics, and traces through search, dashboards, alerts, AI notebooks, and open agent interfaces. ### What does the fully managed experience handle? We handle ingestion, storage, optimization, scaling, backups, and upgrades. You instrument your applications, send telemetry to your managed endpoint, and focus on investigating your systems. ### Do I need an AI agent to use it? No. Engineers can investigate directly through search, trace waterfalls, event patterns, event deltas, dashboards, and session replays. Agents and AI notebooks provide additional ways to explore the same telemetry. ### Can I use my own agents and models? Yes. Connect your preferred agents and frameworks through MCP and SQL. Combine telemetry with the code, runbooks, and internal tools your agents already use. ### How is ClickHouse Observability related to ClickStack? ClickStack is the open source observability stack behind ClickHouse Observability. ClickStack remains available as an open source product, while ClickHouse Observability provides the managed service. ### Why would I choose the self-managed experience? Most teams should choose the fully managed experience, where ClickHouse handles database administration and optimization for you. The self-managed experience suits mature engineering teams with the database expertise and dedicated resources to own configuration, performance tuning, and workload management. It gives you the control to pursue the lowest possible unit costs, while ClickHouse Cloud manages the underlying infrastructure. ### What reliability and security capabilities are included? ClickHouse Cloud provides replication across multiple availability zones, automated backups, and disaster recovery. Security capabilities include user and network access controls, activity logging, encryption in transit and at rest, and SOC 2 Type II compliance. ### How is the fully managed experience priced? Logs, traces, and session replays are billed by uncompressed GB ingested. Metrics are billed per million data points, with no separate charge for cardinality. Both include 30 days of retention, with no monthly minimum. Visit the [pricing page](https://clickhouse.com/pricing/observability) for plan rates and search allowances. ### Does higher metrics cardinality increase my bill? There is no separate charge for the number of unique metric series. Your metrics bill depends on the number of data points you send. At the same data-point volume, the price stays the same regardless of cardinality. Creating additional series can increase costs if it produces more data points. ### How do I get started? [Start a free trial](https://console.clickhouse.cloud/signUp?intent=o11y&loc=clickhouse-observability-faq) and connect your OpenTelemetry SDKs or collectors to your managed endpoint. Then explore your telemetry through the interface or connect your agents.