May 2026 Summaries

ClickHouse celebrated its 10th anniversary as an open-source project during a recent conference, highlighting its rapid ecosystem growth with over 200 integrations and 2,600 contributors. The event introduced the "House Mates" official partner program, including key partnerships with companies like Fivetran and Sigma Computing, which are developing integrations to enhance data replication and ETL processes on ClickHouse Cloud. Significant announcements included the general availability of Fivetran's ClickHouse Cloud connectors and Sigma Computing's public beta for its ClickHouse connector. ClickHouse also partnered with dbt Labs to develop the dbt Fusion adapter, aiming to boost analytics engineering efficiency. Other collaborations were formed with platforms like Apache Airflow, Vercel, and Notion, focusing on seamless data integration and management. Additionally, ClickStack was expanded with AI Notebooks and the MCP server to improve observability workflows. These developments underscore ClickHouse's commitment to providing scalable and innovative data solutions across various domains, from ELT and BI to AI observability.

ClickHouse Open House 2026 Day 1 featured a series of announcements across ClickHouse Cloud, Postgres, distributed query execution, AI agents, observability, and lakehouse interoperability. Key highlights included the public beta release of ClickHouse Postgres, which offers over five times the transactions per second compared to AWS RDS, and significant improvements in query performance through multi-stage distributed queries, reducing TPC-H SF100 execution time nearly by half. ClickHouse Agents, powered by Claude, entered public beta, providing a native chat experience and a no-code agent builder. Enhancements to interoperability included support for writing to Microsoft OneLake and Unity Catalog, enabling bidirectional workflows between ClickHouse and external lakehouse systems. Updates to ClickHouse Cloud aimed at improving resilience, observability, schema management, and developer tools were also unveiled, with features like cross-region replication, Monitoring v2, and a new Schema Management and Optimization tool. The day concluded with the announcement of Langfuse V4, improving query performance by up to 200 times, and the introduction of ClickStack Cloud, a serverless observability platform, alongside a new partner program called House Mates.

ClickHouse announced significant achievements and new initiatives at its second annual user conference, Open House 2026, highlighting a breakthrough quarter with annual recurring revenue surpassing $250 million, a more than threefold increase from the previous year, and a customer base expanding to 4,000 firms. The company introduced "ClickHouse Agents," a fully managed analytics service utilizing Anthropic's Claude AI, allowing users to build and deploy AI agents effortlessly. Additionally, the launch of "CostBench," an open benchmark for comparing the cost performance of major cloud data warehouses, revealed ClickHouse's leading cost efficiency, outperforming competitors by 23 times. The introduction of the first official partner program, "House Mates," aims to strengthen ClickHouse's ecosystem with over 60 global partners and foster deeper integrations with major cloud providers. These developments underscore ClickHouse's strategic focus on enhancing AI-driven analytics, optimizing cost efficiency, and expanding its collaborative network to sustain its competitive edge in the real-time data processing landscape.

ClickHouse Cloud introduces multi-stage distributed execution to enhance query scalability and performance, addressing bottlenecks in large joins and high-cardinality aggregations by repartitioning intermediate data between execution stages. This new approach allows queries to be executed in parallel across multiple nodes, removing limitations of previous models that restricted scalability for modern petabyte-scale workloads. The method employs exchange operators to efficiently move data between stages, optimizing query execution by allowing for dynamic repartitioning and parallel processing. TPC-H benchmark tests demonstrate significant speed improvements, with some queries achieving up to 7.4× faster execution on eight nodes compared to a single node, although certain single-node optimizations are not yet fully supported. The development of a cost-based optimizer is anticipated to further enhance performance by automatically selecting the best aggregation strategies based on various factors. Currently, multi-stage distributed execution is available in an experimental phase within ClickHouse Cloud's private preview program.

The ClickHouse Open House 2026 Day 1 event introduced a series of significant advancements and new features across various domains including ClickHouse Cloud, AI agents, observability, Postgres, and lakehouse interoperability. Key announcements highlighted the public beta release of ClickHouse Postgres, which boasts over 5x more transactions per second than AWS RDS, and multi-stage distributed queries that have halved the TPC-H SF100 runtime. The ClickHouse Cloud introduced updates enhancing resilience, observability, and developer tooling, along with cross-region replication for improved failover capabilities. AI-driven improvements included the public beta of ClickHouse Agents, allowing for a native chat experience and a no-code agent builder, while Langfuse V4 promised a 200x performance boost in query handling. Integration updates also included write support for Microsoft OneLake and Unity Catalog, enabling seamless workflows between ClickHouse and external lakehouse systems. The event also announced the private preview of ClickStack Cloud, a serverless observability platform built on ClickHouse, emphasizing scalability and automatic optimization of telemetry data. Additionally, the introduction of the House Mates partner program aims to foster collaboration and innovation across technology, services, and reseller tracks. Overall, the announcements underscore ClickHouse's commitment to flexibility, performance enhancements, and user control in real-time data systems.

The second day of the ClickHouse Open House conference highlighted the growth and innovation within the ClickHouse ecosystem, celebrating its 10th anniversary as an open-source project. The event showcased various integrations and partnerships, including the launch of the House Mates partner program with over 60 ISV, technology, and service partners. Key announcements included the general availability of Fivetran's connector for data replication into ClickHouse Cloud, the public beta of Sigma Computing's connector, and the development of a dbt Fusion adapter. The native Apache Airflow provider for ClickHouse and Vercel AI SDK v7's integration with Langfuse were also introduced. Notable advancements such as the Python Client v1 and .NET Stack for ClickHouse were revealed, emphasizing performance improvements and expanded compatibility. Additionally, ClickStack's new AI Notebooks and MCP server were unveiled to enhance observability workflows, demonstrating the dynamic evolution and collaborative spirit of the ClickHouse community and its partners.

Alexey Milovidov, the creator of ClickHouse, embarked on an ambitious 12-day tour across five countries in the Asia-Pacific and Japan (APJ) region, which included stops in Mumbai, Bengaluru, Singapore, Jakarta, Seoul, and Tokyo. During this whirlwind journey, he engaged in various activities such as customer meetings, keynotes, executive roundtables, press interviews, and meetups, all aimed at strengthening ClickHouse's presence in the region. The trip highlighted the importance of face-to-face interaction in fostering a deeper understanding of customer needs and receiving real-time feedback, which is vital for improving the ClickHouse database. This tour also culminated in the official launch of ClickHouse in Japan, marking a significant milestone for the company. The initiative underscored the commitment to building a community-driven product by engaging directly with users and partners, further solidifying ClickHouse's market position in one of its fastest-growing regions.

ClickHouse Cloud now offers a fully managed Postgres service, leveraging local NVMe storage for enhanced transaction speeds and integrating seamlessly with ClickHouse for real-time analytics. This unified data platform combines Postgres for OLTP and ClickHouse for OLAP, simplifying the development of AI-native applications by eliminating the need for complex system integrations. The service, free until June 15, 2026, and discounted during the beta phase, includes features like native CDC for data synchronization, a unified query layer with pg_clickhouse, and managed migration workflows through ClickPipes. It caters to a wide range of industries, including cybersecurity, fintech, and retail, with enterprise-grade capabilities such as high availability, security, and extensive PostgreSQL extensions. The platform's flexibility supports a variety of workloads, from lightweight to storage-intensive deployments, and aims to provide a cost-effective solution for building reliable, real-time data applications.

CostBench is an open benchmark designed to evaluate the cost-performance of cloud data warehouses, focusing on performance-per-dollar rather than just speed to help teams choose the most efficient system for real-time analytical workloads. It emphasizes that speed alone is insufficient because costs vary based on how much compute power a system requires to complete tasks, with different platforms using various units to measure this. CostBench specifically measures both read-side cost-performance, which assesses query efficiency per dollar, and write-side cost-performance, which evaluates the cost-effectiveness of processing fresh data for queries. The initial release highlights read-side performance across platforms like ClickHouse Cloud, Snowflake, Databricks, BigQuery, and Redshift, revealing ClickHouse Cloud as the only option maintaining fast and low-cost performance as data scales. By publishing its methodologies, scripts, and results openly, CostBench allows users to inspect, verify, and potentially improve the configurations, removing the opacity often associated with cost-performance assessments.

ClickStack Cloud is a fully managed observability service built on ClickHouse that offers teams the ability to use ClickHouse for monitoring without managing the infrastructure. It allows users to send OpenTelemetry data to a managed endpoint, where ClickStack Cloud handles ingestion, buffering, scaling, and storage automatically. This service is designed for teams seeking the performance and cost-efficiency of ClickHouse for observability without the complexity of infrastructure management, enabling them to explore logs, metrics, and traces in the ClickStack UI seamlessly. During its private preview, ClickStack Cloud is focused on automatic schema tuning and dedicated query compute to optimize performance for high-volume observability workloads. It aims to provide a simplified user experience by managing the operational complexity behind the scenes, making it suitable for teams that want to leverage ClickHouse's power for observability tasks like incident correlation and data retention without becoming experts in ClickHouse operations.

Open House, a recent event focused on the ClickHouse community, introduced significant updates aimed at enhancing real-time data observability. Key announcements included the launch of ClickStack Cloud in private preview, offering a serverless, fully managed platform for exploring logs, metrics, and traces without the need for direct infrastructure management, and Managed ClickStack, which provides teams with operational control over observability stacks. Additionally, AI Notebooks, now in beta for Managed ClickStack, were designed as collaborative investigative workspaces, allowing engineers to explore complex incident investigations with persistent, structured workflows. The event also revealed the ClickStack MCP server, enabling external AI systems to integrate with ClickStack's observability tools, thus supporting diverse operational models. These developments underscore a shift towards more collaborative and programmable observability systems, allowing engineers to investigate systems flexibly without being confined to predefined workflows.

ClickHouse has demonstrated significant growth and innovation, with its annual run-rate revenue surpassing $250 million, more than tripling from the previous year, and expanding its customer base to over 4,000. The company announced several new initiatives at its Open House 2026, including ClickHouse Agents, a fully managed agentic analytics service powered by Anthropic's Claude, and the CostBench benchmark, which evaluates cloud data warehouses on cost-performance, showing ClickHouse Cloud as 23 times better than its nearest competitor. Additionally, ClickHouse introduced "House Mates," its first formal partner program, comprising over 60 technology and service partners designed to enhance integration and market reach. These developments underscore ClickHouse's commitment to addressing the demands of AI-era workloads with high-performance and cost-efficient data infrastructure solutions, positioning the company as a leader in the analytics market.

Over 500 data engineers gathered at ClickHouse's annual user conference in San Francisco to celebrate significant milestones, including surpassing 4,000 customers and achieving $250 million in annual run-rate revenue. Originally created by Alexey Milovidov and open-sourced in 2016, ClickHouse has grown into a standard for real-time analytics, supported by a decade of community contributions. The conference highlighted the diverse applications of ClickHouse by companies such as Anthropic, Capital One, and Tesla, and introduced new offerings like ClickHouse Agents, a no-code agentic analytics service, and CostBench, an open benchmarking tool. The announcement of House Mates, a formal partner program, and various new features such as Managed Postgres and observability tools, underscored ClickHouse's commitment to innovation and efficiency. The event emphasized the importance of cost-performance in AI adoption and expressed the company's forward-looking vision to continue evolving for future customers.

House Mates is a partner community and program created by ClickHouse to enhance customer experiences by providing over 60 partners, including ISVs like dbt Labs and Grafana Labs, and consulting firms such as TigerData and DoIT, trained through ClickHouse Academy. The program is structured into three tiers—Ignite, Accelerate, and Prime—offering a framework for partners to collaborate and innovate, benefiting from joint go-to-market strategies and attractive incentives. This initiative addresses the need for validated integrations and certified experts, supporting ClickHouse's rapid global adoption, with over 4,000 customers using its services for real-time analytics and other demanding workloads. House Mates aims to connect various technologies and expertise, facilitating seamless integration and modernization of data platforms, as demonstrated by successful collaborations in regions like LATAM, APJ, and the Middle East. The program is committed to evolving by expanding its partner network and refining its approach based on feedback to ensure reliable, scalable solutions for both customers and partners.

ClickHouse's native random sampling feature allows users to execute queries on a fraction of their data, providing faster query times while maintaining a reasonable level of accuracy. By using the UK house prices dataset with over 30 million transactions, the process involves creating a table with a suitable sample key, such as the sipHash64 function applied to high-cardinality columns like postcode combinations, to ensure an even distribution of the sampled data. The approach demonstrates how to leverage sampling for both fractional and row count-based queries, highlighting the benefits of reduced processing time and resource usage. To optimize results, the sampling key should be included at the beginning of the ORDER BY clause, and sum aggregations should be scaled using the _sample_factor virtual column. This method is particularly effective for exploratory data analysis where approximate answers are sufficient, offering an efficient trade-off between accuracy and performance.

ClickHouse has advanced its capabilities to support open table formats like Delta Lake by integrating the Rust Delta Kernel, simplifying the process of managing complex table protocols. This integration allows ClickHouse to handle transactional data, schema evolution, and time travel with ease, without the burden of implementing these features natively. The Rust Delta Kernel abstracts the underlying complexities of Delta Lake, providing a consistent interface that reduces maintenance overhead while enabling high-performance query execution. This approach enhances ClickHouse's ability to execute analytical workloads efficiently, leveraging Delta Lake's capabilities for transactional consistency, schema evolution, and partition pruning. By contributing to the Delta Kernel's development, ClickHouse aims to improve the broader data ecosystem, ensuring seamless operation across various analytical needs. The integration reflects a strategic shift towards shared abstractions in data management, allowing ClickHouse to focus on optimizing performance while maintaining robust support for evolving data protocols.

The May 2026 ClickHouse newsletter highlights significant advancements in observability and AI integration, showcasing how companies like Qonto and LINE MAN Wongnai have utilized ClickHouse Cloud for enhanced performance and efficiency. Javier Ortiz, the featured community member, led Qonto's migration from Grafana Tempo to ClickHouse Cloud, drastically reducing data storage needs and expanding query capabilities. The newsletter also covers various topics such as agentic analytics in financial services and the advantages of ClickHouse over Elasticsearch for log analytics. Additionally, it previews the upcoming Open House 2026 conference, focusing on ClickHouse's real-time analytics and database administration, and discusses the latest 26.4 release, which includes SQL compatibility improvements and new functions. The newsletter encourages engagement through community events and training sessions, emphasizing the role of ClickHouse in transforming data processing and analytics across industries.

The release of clickhousectl v0.2.0 introduces several new features, including support for managing Postgres both locally and in ClickHouse Cloud, ClickPipes management for a range of data sources, and the ability to execute SQL over HTTP using ClickHouse Cloud's Query Endpoints without requiring a local ClickHouse binary. This update also adds an agent-friendly output format using Markdown-style tables and a standalone Rust client library for the ClickHouse Cloud API. Users can now manage Postgres instances with high availability, read replicas, and point-in-time restore capabilities, while ClickPipes offers managed connectors for streaming and batch data ingestion from various sources like S3, Kafka, and BigQuery. The update aims to enhance user experience with improvements such as debug options for credential resolution and is part of a broader effort to optimize for agentic experience (AX) in the future. The tool, still in its early stages, has quickly gained traction with over 5000 developers, and the team is actively working on expanding features and gathering community feedback.

In April's update for ClickStack, significant improvements were made to enhance the core experience of querying, alerting, and dashboards. The introduction of SQL-powered alerts allows users to seamlessly transition from writing queries and building dashboards to setting alerts without switching languages, enhancing the SQL-native observability workflow. Performance optimizations, including a redesigned logs schema and a major overhaul of autocomplete leveraging materialized-view rollups, offer faster and more responsive data handling. Heatmaps have been elevated to a first-class chart type, enabling easier visualization of data distributions within dashboards. Additional quality-of-life improvements include more flexible alert thresholds, enhanced alert history tracking, and user-friendly features such as scrollable pie chart legends and per-series number formatting. April's release also emphasizes community involvement through open source contributions and offers a preview of upcoming observability talks at the Open House event, showcasing how various teams utilize ClickStack with ClickHouse in production.

ChatFeatured, a Toronto-based startup, helps brands influence their visibility in AI search engines by optimizing content and analytics. Initially relying on Postgres for both application and analytics needs, they faced performance issues as their customer base grew. To address this, they combined Postgres for transactional tasks and ClickHouse for analytical queries, significantly improving speed and efficiency. This integration allowed them to cut analytics query times from minutes to milliseconds, enhancing user experience and enabling new features. The switch was facilitated by ClickHouse's managed Postgres service, which offers a seamless data sync between the two databases. This move not only improved performance but also reduced operational overhead, providing a unified platform that supports their rapid growth and customer demands. ChatFeatured's co-founder, Nithiiyan Skhanthan, highlights the importance of execution and speed in delivering a superior user experience, which has attracted global customers, including billion-dollar companies, and positioned the startup as a leading solution in the AI-driven brand visibility space.

The text delves into the challenges and solutions associated with handling high cardinality in observability data, contrasting the approaches of column-oriented databases like ClickHouse with series-oriented systems such as Prometheus. While Prometheus faces complexities with its model of treating each unique label combination as an independent series, resulting in memory overhead and operational complexity, ClickHouse leverages a "wide events" approach. This method represents metrics as rows with attributes that can be aggregated at query time, reducing the strain of high cardinality by storing data in a columnar format with compressed, sharded maps. While ClickHouse excels in flexibility, allowing dynamic labels and reducing I/O through efficient filtering and compression, it is not without its challenges, such as query-time costs for large GROUP BY operations. The text emphasizes that ClickHouse is not a direct replacement for Prometheus, highlighting its strengths in handling event-style observability data and long-term trend analysis, while Prometheus remains effective in scenarios requiring strict time-series semantics and real-time alerting.

The D. E. Shaw group, a global investment and technology development firm, employs ClickHouse to enhance high-cardinality observability across millions of compute workloads on its internal grid, significantly outperforming alternatives by a factor of seven in evaluations. This transition allowed the firm to handle over 500,000 records per second in ingestion and support long-term capacity planning and strategic decision-making. By maintaining a unique task ID for each workload, they capture telemetry data crucial for understanding compute usage at a granular level, although this approach introduces high cardinality that their existing platform couldn't manage. The implementation of ClickHouse provides improved compression and query performance, enabling the firm to expand observability into distributed tracing and other event-data pipelines. The shift involved translating Prometheus-style queries into SQL, presenting initial challenges but ultimately proving advantageous. ClickHouse's performance, particularly in data ingestion and query efficiency, supports detailed analysis and capacity planning, assisting the firm in making informed decisions regarding compute resource allocation. As the firm continues to explore broader event-data pipelines and structured logs, ClickHouse's role as a unified observability platform grows, demonstrating its capability to handle high-volume, high-cardinality data with the right design and analytical foundation.

The integration of AI demands a reimagined data strategy to support high concurrency, real-time query processing, and full-fidelity data, challenging traditional batch-oriented architectures. The convergence of previously siloed use cases, such as data warehouses and observability, is underway, with platforms like ClickHouse evolving to support AI workloads across applications, analytics, and SRE experiences. The shift towards agent-driven applications, conversational analytics interfaces, and AI-driven observability requires a unified platform that can handle both transactional and analytical workloads efficiently. The combination of Postgres for transactional data and ClickHouse for analytics is becoming the modern data stack for data-intensive applications, while legacy data warehouse architectures struggle to meet the demands of AI analysts and their high-frequency, low-latency query needs. The observability domain is also transitioning from a metric, log, and trace model to a full-fidelity event model, driven by AI SRE workflows that demand granular, high-cardinality data. Platforms like ClickHouse, which offer efficient data handling and cost models based on compute and storage, are positioned to support these changes, while legacy systems face the challenge of adapting to this new paradigm. Overall, the future of data platforms involves integrating AI-native tools and observability solutions to support interactive AI-driven applications.

High cardinality poses significant challenges in traditional time-series systems like Prometheus due to the creation of numerous unique time series, each carrying overhead in terms of memory, indexing, and lifecycle management. In observability workloads, this complexity arises from the many unique label combinations that define metrics, leading to increased memory consumption and potential system instability, particularly in environments with high churn or ephemeral workloads. Prometheus, while efficient for moderate numbers of long-lived series with regular intervals, struggles with high cardinality as it leads to memory pressure and necessitates compromises such as limiting labels or reducing metric resolution to manage cardinality explosion. This impacts both write-time performance, where each new series incurs overhead, and read-time performance, where broad queries across high-cardinality dimensions require extensive data processing. These constraints often force users to trade off between visibility and system stability, highlighting the need for alternative approaches, like ClickHouse, which handle such workloads differently through column-oriented storage and other optimizations.

Postgres extensions enhance functionality by allowing users to integrate features not originally included in Postgres, such as geospatial data handling with PostGIS and time-series data management with TimescaleDB. Foreign Data Wrapper (FDW) extensions enable Postgres to interact with external data sources, like ClickHouse, by creating foreign tables, allowing SQL queries across both systems. The process of determining what parts of a query are executed remotely, known as pushdown, is complex, involving the translation of SQL expressions between Postgres and ClickHouse to ensure accurate results. The development of pg_clickhouse, a ClickHouse FDW, involves iterative enhancements to expand pushdown capabilities for various SQL clauses such as window functions and JSON operations, significantly reducing data transfer volumes and execution times. The goal is to maximize remote execution while maintaining the integrity of query results, highlighting the intricate negotiation between different SQL grammars to optimize performance in a unified data stack.

ClickStack has introduced SQL-based visualizations and alerting, enabling users to create advanced charts and alerts using ClickHouse SQL queries within the ClickStack UI. This development offers greater flexibility compared to traditional query builders, which often limit advanced users. SQL-based tools allow for complex analyses like rolling averages, anomaly detection, and grouped statistical checks, directly enhancing observability workflows. The system supports dynamic, interactive dashboards through query parameters and macros, facilitating seamless integration with existing tools like Grafana. SQL-driven alerting, an extension of charting, allows detailed operational patterns to be expressed directly within queries, simplifying alert configurations by encapsulating logic within the SQL itself. This capability is particularly beneficial for defining alerts based on statistical analysis rather than static thresholds, thereby broadening the range of alerting strategies available to users.

In ClickHouse, dictionaries and Join tables are utilized to optimize join performance in dimensional modeling, a technique involving facts and dimensions based on the Kimball methodology. While dictionaries hold dimension data in memory for direct joins, the Join table engine allows for in-memory structures supporting specific join types, enhancing performance by persisting data on disk. Despite some drawbacks in the open-source version, such as lack of distribution and inefficient handling of frequent updates, ClickHouse Cloud addresses these issues with a SharedJoin table backed by a MergeTree family, enabling efficient upserting, deduplication, and data compaction. This structure is especially beneficial for implementing Type 1 slowly changing dimensions using an ANY LEFT join, ensuring that only the latest entries are maintained in memory. The cloud version thus facilitates scalable and high-performing data enrichment processes.

Avride leverages ClickHouse Cloud as the central data infrastructure for its autonomous vehicle and delivery robot fleet, significantly improving data processing efficiency after transitioning from Apache Iceberg. This migration resulted in a dramatic reduction of index lookup latency from 20 seconds to under 100 milliseconds and ingestion latency from hours or days to mere seconds. Despite the team's extensive experience with ClickHouse, they opted for ClickHouse Cloud due to its operational simplicity and the benefits of separating storage and compute, which allows for scalable and cost-effective data management. Their data pipeline begins with vehicles generating vast amounts of sensor data, which is processed to produce both offline and online metrics for various use cases, from real-time fleet monitoring to C++ performance profiling. The choice to use ClickHouse Cloud is driven by its ability to handle parallel writes efficiently and eliminate data duplication, which were limitations with Iceberg. ClickHouse Cloud's features, including its SQL console, enhance usability across the company, supporting a wide range of applications from engineering to executive decision-making. As Avride's fleet and data demands grow, ClickHouse Cloud provides a robust and scalable solution, making it their default choice for future expansion.

The ClickHouse 26.4 release introduces numerous enhancements, including 39 new features, 45 performance optimizations, and 238 bug fixes, furthering its SQL compatibility and improving various functionalities. Notable improvements include a faster COUNT DISTINCT operation, a more user-friendly EXPLAIN function, and enhanced SQL features such as the VALUES table expression, EXTRACT operator with PostgreSQL-style units, and SET TIME ZONE. The release also highlights an optimized LIKE query using text indexing, leading to significant performance gains by employing an inverted index to reduce full-table scans, and the addition of JSONAllValues to improve filtering on JSON subcolumns through text indexing. The community has welcomed new contributors, and users are encouraged to explore these updates and get started with ClickHouse Cloud, which offers a $300 credit for new users.

Query Insights is a diagnostic tool in preview for ClickHouse Cloud Managed Postgres that helps identify and address slow query patterns by providing detailed telemetry data for every query executed on a database. It utilizes the open-source extension pg_stat_ch to stream per-statement telemetry into ClickHouse, allowing users to access an overview of database health, identify slow query patterns, and analyze specific execution details to determine causes of latency, such as disk spills during sorting or insufficient parallel workers. The tool is designed to integrate with existing systems by using ClickHouse Cloud for data storage and query, ensuring efficient handling of large volumes of telemetry data while maintaining the privacy of sensitive information. Future developments for Query Insights include an open API, wait event attributions, EXPLAIN plan exposures for slower queries, and actionable recommendations to further enhance the user experience. The tool aims to streamline performance troubleshooting by offering clear insights into database operations and actionable steps for optimization.

Agentic workloads, characterized by their continuous and high-concurrency demands, challenge traditional analytical systems by requiring fresh and query-ready data at a low cost. ClickHouse emerges as a more cost-effective solution compared to Snowflake by integrating data ordering directly into the write path, which results in a 22× lower cost for obtaining query-ready data and a 28× better write-side cost-performance. This architectural difference allows ClickHouse to efficiently manage data storage and retrieval without relying on a separate clustering process, unlike Snowflake, which clusters data post-ingest, incurring additional costs. ClickHouse's approach not only ensures immediate query-readiness but also leads to better data compression and lower storage costs over time. Such efficiency is crucial in the agentic era where systems must handle continuous data ingestion and provide rapid, complex insights, positioning ClickHouse as a cost-efficient choice for real-time analytics at scale.

Outages in software systems often originate in observability gaps, where missing spans and broken traces leave engineers guessing about failures, even in systems that seem fully instrumented. While modern observability tools generate vast amounts of data, this doesn't necessarily translate into better insights without complete coverage and contextual understanding. The combination of ClickStack, a ClickHouse-native observability backend, and Odigos, a zero-code OpenTelemetry instrumentation platform, addresses these challenges by offering high-quality telemetry capture across distributed systems. Odigos employs eBPF for dynamic, non-intrusive instrumentation, automatically generating comprehensive traces, including application-specific metadata and asynchronous messaging flows, without modifying application code. ClickStack complements this by providing a scalable backend for ingesting and querying this enriched telemetry, enabling engineers to efficiently identify issues by offering a complete, context-rich view of system behavior. Together, these tools transform observability into a precise, data-driven process, eliminating guesswork and enhancing system reliability.

Gala, a blockchain-powered platform offering games and media, significantly enhanced its data analytics capabilities and reduced costs by migrating from its previous data infrastructure to ClickHouse Cloud on AWS. This transition enabled Gala to handle a growing volume of data efficiently, drastically reducing query times from minutes to sub-seconds and increasing data storage from 3 TB to 9 TB. The move allowed Gala's engineering teams to focus more on strategic initiatives rather than data management, while also empowering non-technical employees to conduct their own analytics through Metabase, a user-friendly BI tool. With the support of ClickHouse, Gala experienced a 30% reduction in costs and improved system reliability, freeing its engineers from frequent maintenance issues. This strategic shift not only provided immediate performance benefits but also laid the groundwork for future enhancements using ClickHouse’s ClickPipes data-processing pipeline, ensuring that Gala's data infrastructure can grow alongside its expanding user base and product offerings.

Qonto, a digital bank serving over 600,000 small businesses and freelancers in eight European countries, transitioned to ClickHouse Cloud to enhance observability across its banking platform, switching from Grafana Tempo to address limitations in querying and data compression. ClickHouse's ability to extend query windows and compress data significantly reduced storage costs and enabled real-time insights without sampling constraints. This shift allowed Qonto to adopt a more flexible approach to incident investigation, leveraging an AI companion powered by the ClickHouse MCP server to analyze incidents in plain English, leading to more efficient and democratized incident response. The integration of ClickHouse facilitated collaboration between observability and data engineering teams, enabling them to explore real-time data processing and pre-aggregation pipelines. This move marked a philosophical shift in Qonto's approach to observability, treating it as a holistic data problem and opening up new avenues for product teams to understand user behavior and feature adoption, all while reducing the need for protecting infrastructure and empowering teams to innovate using ClickHouse's capabilities.