November 2025 Summaries

Understanding the real costs of cloud data warehouses requires looking beyond public price lists, as each platform uses different billing units and execution models that affect compute costs. The five major cloud data warehouses—Snowflake, Databricks, ClickHouse Cloud, Google BigQuery, and Amazon Redshift Serverless—each have distinct ways of billing for compute resources. Snowflake and Databricks use a provisioned-cluster model, where users pay for the fixed resources of a cluster, while ClickHouse Cloud allows more flexible node sizing with a normalized billing unit. In contrast, BigQuery and Redshift Serverless employ a serverless approach, billing based on actual compute consumption rather than pre-provisioned resources. This analysis uses the Bench2Cost framework to compare these systems by translating benchmark runtimes into costs, emphasizing that billing models, rather than query runtimes, dictate the cost. The analysis concludes that ClickHouse Cloud offers superior cost-performance for analytical queries due to its innovative billing and scaling models.

In a comprehensive benchmark comparing major cloud data warehouses—Snowflake, Databricks, ClickHouse Cloud, BigQuery, and Redshift—across datasets of 1 billion, 10 billion, and 100 billion rows, ClickHouse Cloud consistently provided superior cost-performance. The study utilized Bench2Cost, an open-source benchmarking pipeline, to ensure accurate cost comparisons by applying each vendor's real compute billing rules. While storage costs were analyzed, the focus remained on compute costs due to their greater impact on analytical workloads. The analysis revealed that as data scales, most systems drift towards higher costs and slower performance, with ClickHouse Cloud standing out as the only system maintaining efficiency and low cost across all scales. The findings are supported by an interactive benchmark explorer, which allows users to delve into the detailed results and configurations. The study highlights ClickHouse Cloud's resilience and cost-effectiveness, especially at larger data scales, contrasting with the limitations faced by other platforms such as Snowflake and Databricks at their maximum configurations.

ClickStack leverages event patterns to enhance log and event analysis by clustering similar logs and traces, allowing users to identify key behaviors and anomalies without predefined rules. Built on the Drain3 algorithm, ClickStack performs dynamic pattern analysis directly in the browser, enabling engineers to manage large volumes of unstructured data more efficiently. The system mitigates the challenge of dissecting logs by automatically detecting and grouping recurring structures, offering a high-level overview of data and surfacing both frequent patterns, which indicate normal behavior, and rare patterns, which may reveal anomalies. This approach complements traditional search methods and aids in improving observability by helping teams identify redundant logs and improve logging practices. Event patterns are integrated at query time, avoiding the computational expense of clustering at ingestion, and using a sampling strategy to maintain efficiency and accuracy. ClickStack's use of Pyodide allows it to execute Python code directly in the browser for seamless interaction and visualization, providing a practical solution for navigating complex data landscapes and setting the stage for future advancements in analytics and data compression.

Mike Heffner and Ray Jenkins from Streamfold explore the efficiency improvements achieved with Rotel, an open-source Rust project designed for high-performance OpenTelemetry data collection, in comparison to the OpenTelemetry Collector. By constructing a benchmarking pipeline to stream trace data into ClickHouse, they demonstrate how Rotel scales throughput from 1.1 million trace spans per second to 3.7 million on the same hardware, achieving nearly four times the efficiency of the OpenTelemetry Collector. This is accomplished through various optimization techniques, such as JSON binary serialization, improved task management, and leveraging LZ4 compression, as well as addressing allocator lock contention that significantly reduced CPU usage. The authors stress the importance of efficiency at scale, especially in large systems, and highlight future areas of exploration including enhancements to Rotel's handling of logs and metrics, as well as further optimization of memory allocation impacts. The post emphasizes the potential for significant resource savings at large scales, positioning Rotel as a highly efficient solution for streaming OpenTelemetry data into ClickHouse.

Canva has significantly enhanced its real-time observability infrastructure by leveraging ClickHouse, achieving a 14x compression ratio and reducing infrastructure costs by around 70%. By redesigning data ingestion and storage, Canva now processes nearly 3 million spans and logs per second, which is critical given their 240 million monthly active users. This transformation has led to a 10x performance improvement in trace search times, reducing them from 30 seconds to 2.5 seconds. The journey involved moving tracing workloads to ClickHouse, optimizing ingestion, and refining schema design to support scalable and efficient data handling. For trace searches, Canva implemented a materialized-view lookup table and client-side optimizations to enhance speed, while their logging infrastructure benefited from async inserts for improved performance. The team also addressed the developer experience by building a custom query gateway and logging platform, ensuring a familiar and efficient user experience. This overhaul has resulted in a leaner and faster observability platform, with Canva's traces and logs running on ClickHouse with fewer resources than their previous system required.

FabFunnel, developed by IdeaClan, is a digital marketing platform designed to streamline campaign management by integrating creative workflows, automation, and reporting from a single dashboard, with real-time data integration from major ad platforms like Facebook, TikTok, and Google. Initially launched at Affiliate World Budapest in 2025, the platform faced challenges with its MySQL-based architecture, which struggled to handle the high volume of real-time data necessary for efficient marketing analytics. To address these issues, FabFunnel transitioned to ClickHouse Cloud, a decision that significantly enhanced system performance by enabling fast, high-volume data inserts and sub-second query responses, thanks to features like Confluent Kafka for data ingestion, ReplacingMergeTree tables for real-time updates, and refreshable materialized views for efficient data aggregation. This migration not only improved query efficiency and reduced infrastructure overhead but also allowed FabFunnel to deliver instant analytics, enabling media buyers to manage and scale campaigns effectively without infrastructure constraints.

The AWS MCP Hackathon in San Francisco showcased innovative AI agents that leverage real-time data for applications in adtech, AI operations, and digital health, with ClickHouse as the analytical backbone. The event highlighted three standout projects: an Ad Optimizer Agent for predicting ad bidding prices using real-time signals, an AI Ops Agent that acts as an incident commander for reliability in API and model failures, and a GlucoTrack Predictive Agent for managing Type 1 diabetes through glucose monitoring and forecasting. These projects demonstrated the value of ClickHouse's low-latency, scalable analytics, enabling instantaneous data processing and interaction in user interfaces. The hackathon illustrated the power of combining streaming data technologies like Kafka and orchestration tools like Temporal, emphasizing the importance of efficient data handling and real-time analytics in developing robust AI-driven solutions. ClickHouse expressed enthusiasm for future collaborations and hackathons, recognizing its platform's potential to drive advancements in real-time, data-driven innovations.

ClickHouse Inc. has announced a new integration with Microsoft OneLake, enhancing their collaboration to deliver real-time, large-scale analytical capabilities by combining the speed of ClickHouse with the scalability and governance of Microsoft's unified data lake, OneLake. This integration allows organizations to conduct federated and unified analytics, supporting various workloads, including real-time analytics, operational dashboards, and AI use cases, by exposing Iceberg tables to ClickHouse users. The collaboration aims to simplify data ingestion and analysis, enabling Microsoft customers to leverage ClickHouse's high-performance database management system for data-intensive applications such as real-time observability and cybersecurity. ClickHouse is recognized for its fast, open-source columnar database management system, trusted by companies like Sony, Tesla, and Lyft, and is engineered to handle massive volumes of data with exceptional query speed and concurrency.

ClickHouse has announced a new integration with Microsoft OneLake, part of Microsoft Fabric's unified data lake, which will be available as a beta feature in ClickHouse 25.11 and later in ClickHouse Cloud. This integration allows ClickHouse to directly query Iceberg tables in OneLake, facilitating seamless and efficient analytics across various data sources. Microsoft OneLake enables organizations to unify their data into a single repository, cataloging and securing it for analytics and AI projects, with compatibility for external stores like Amazon S3 and Google Cloud Storage. The integration leverages OneLake’s Iceberg-compatible Tables APIs, allowing users to query data with minimal configuration while ensuring secure access through Microsoft Entra ID. ClickHouse users can deploy their services on Azure and connect to the OneLake Iceberg catalog, enabling complex analytics and AI initiatives. Future enhancements include moving the feature to general availability, adding write support, and improving cloud integration with a user-friendly interface in ClickHouse Cloud.

Reco, an AI-driven SaaS security solution, focuses on securing enterprises by delving into SaaS applications where employees operate, supporting over 215 integrations and capable of discovering more than 60,000 applications. Initially reliant on Postgres, Reco switched to ClickHouse Cloud to manage its vast data processing needs, ingesting billions of events weekly and handling extensive row reads and writes. The transition to ClickHouse was driven by the need for faster, scalable processing, as Postgres struggled with the volume and complexity of Reco's operations. Despite not being ACID compliant, ClickHouse offered the relational-style functionality required, with advantages such as incremental materialized views, efficient joins, and data-skipping indexes, which improved performance and reduced costs. Additionally, Reco creatively employed ClickHouse as a state store for streaming workloads, overcoming earlier memory and synchronization challenges with Spark and Redis. The switch to ClickHouse Cloud has markedly enhanced Reco's platform, reducing costs, increasing speed, and allowing for significant customer base growth.

chDB is an embedded OLAP SQL engine that integrates ClickHouse's analytical capabilities into Python, eliminating the need for a standalone ClickHouse server. The recent upgrade to the ClickHouse kernel version v25.8.2.29 introduced new features and performance improvements, although it also posed technical challenges. chDB's architecture allows for zero-copy data transfers and supports multiple data formats, and it utilizes two dynamic libraries to address multi-Python version compatibility issues. Memory management presented a significant challenge, particularly with memory allocation and deallocation conflicts between jemalloc and glibc, which was resolved through runtime memory source detection. The upgrade improved performance, notably in complex queries, by optimizing memory management and reducing lock contention, and enhanced compatibility with ClickHouse's latest features. Additionally, the improvements were crucial in reducing overhead and execution times, making chDB more competitive with ClickHouse-local in terms of performance for both single and batch queries. Overall, the kernel upgrade significantly boosted chDB's efficiency and broadened its functionality while maintaining compatibility with ClickHouse's core enhancements.

The November 2025 ClickHouse newsletter highlights significant developments, including the acquisition of LibreChat to create the "Agentic Data Stack," which combines LibreChat's multi-LLM interface with ClickHouse's analytical capabilities to allow users to query data in plain English. The newsletter also features community member Kiyose Ryu, who contributed to building a real-time aggregation system at Smartnews using ClickHouse. Additionally, recent updates in ClickHouse 25.10 introduce various JOIN performance enhancements and new SQL syntax features. The newsletter covers ClickHouse's global community events, including virtual training sessions on real-time analytics and observability at scale, and emphasizes a strategic partnership with Japan Cloud to establish ClickHouse K.K. in Japan. It also explores various technical advancements, such as a comprehensive monitoring system for asynchronous inserts, a vector search engine with precision control, and a low-code data warehouse built with ClickHouse Cloud, while offering insights into projects like achieving 170x log compression and integrating ClickHouse with Salesforce.

Merkle Science, an AI-powered blockchain analytics and predictive risk platform, helps various entities combat illicit activity in the expanding Web3 ecosystem by effectively managing vast amounts of on-chain data. Initially, they relied on a self-hosted ClickHouse setup to handle both OLAP and OLTP functions but found it cumbersome due to infrastructure demands and scaling challenges. In March 2024, the team transitioned to ClickHouse Cloud, alleviating the operational overhead and enhancing performance without compromising speed, which allowed them to focus on feature development and quickly adapt to evolving threats. The migration resulted in significant improvements in data ingestion processes, developer productivity, and system monitoring, enabling the team to innovate and scale seamlessly as blockchain adoption grows. With the cloud infrastructure in place, Merkle Science is positioned to manage up to 130 TB of compressed data, running over 18 million queries weekly, and is prepared to handle further increases in data volume without concern.

StockHouse is a demo application designed to showcase the creation of a streaming market analytics app by leveraging technologies such as ClickHouse, Massive, and Perspective. It functions by ingesting live market data from Massive's WebSocket APIs, storing it efficiently in ClickHouse, and visualizing it in real-time through a web dashboard that updates within milliseconds. StockHouse exemplifies the handling of high-frequency, high-volume data through a streamlined pipeline consisting of multiple components, including a Go-based ingester, a Node.js backend, and a Vue.js frontend utilizing Perspective for visualization. This architecture enables the system to process millions of events per second while maintaining low latency, making it suitable for building trading dashboards, analytics tools, or observability systems. By demonstrating an effective integration of a real-time data source, an analytical database, and a visualization layer, StockHouse serves as a reference architecture for those interested in developing real-time analytics systems.

ClickHouse's internal data warehouse has undergone a significant transformation from a traditional BI-first approach to an AI-first analytics platform, addressing challenges faced by various teams in accessing and analyzing data. Initially, the traditional workflow required substantial SQL expertise, causing delays and bottlenecks in obtaining insights. However, advancements in AI, particularly with Anthropic's Claude models and the Model Context Protocol (MCP), have enabled the creation of DWAINE (Data Warehouse AI Natural Expert), an AI assistant that simplifies data exploration and empowers non-technical users to obtain insights without SQL knowledge. DWAINE's integration of a comprehensive business glossary and enhanced data quality processes has democratized data access, reducing the workload on the data warehouse team by 50-70% and improving efficiency across sales, finance, and operations. Despite the success, approximately 30% of analytics tasks remain in traditional BI tools for repeatable metrics, certified financial metrics, and specialized analyses, ensuring critical decisions are validated through traditional methods. The shift to AI-first analytics has not only improved operational efficiency but also transformed the data culture at ClickHouse, making analytics more accessible and integrated into decision-making processes.

OpenTelemetry (OTel) is transforming observability by breaking away from proprietary systems that limit flexibility and control over telemetry data. By using community-standard instrumentation and the OTel Collector, organizations can freely manage their metrics, logs, and traces, facilitating seamless integration with various tools without being locked into a single vendor. This shift allows for the implementation of SODA (Send Observability Data Anywhere), which enables deliberate data routing to the most suitable tools, enhancing cost efficiency, flexibility, and innovation in observability platforms. OTel supports a pluggable architecture, allowing organizations to adapt their data transport as needs evolve and store historical data in open formats for long-term access and analysis. This approach not only reduces costs but also enables leveraging telemetry data for broader business insights, bridging the gap between observability and internal analytics. While the flexibility offered by OTel and SODA increases opportunities for innovation and efficiency, it also requires careful management to avoid fragmented data silos that can complicate incident response and analysis.

Postgres ClickPipes has introduced a feature allowing the creation of logical replication slots with failover enabled, enhancing high availability (HA) for Change Data Capture (CDC) workflows by preserving replication slots post-failover. This development follows the introduction of logical decoding capabilities in hot standby replicas in Postgres 16 and the subsequent logical replication failover support in Postgres 17. This feature is critical for users who rely on Postgres as a core database for transaction processing and wish to maintain HA while replicating data to other sources for diverse query workloads. Although the ClickPipe toggle simplifies enabling failover, the setup process involves configuring settings to ensure synchronization between primary and standby servers, especially for users employing managed services like AWS RDS or Google CloudSQL. This advancement enables users to leverage ClickHouse's performance for real-time analytics while maintaining data reliability, with self-hosted users advised to consider PeerDB for robust CDC support.

Hud, founded in 2023, introduces itself as the world's first runtime code sensor designed to transform observability by dynamically gathering context on code behavior at both the function and business levels in live production environments. Unlike traditional observability tools that rely heavily on manual logs and traces, Hud focuses on efficiently capturing and analyzing function-level telemetry to quickly identify and resolve issues, minimizing data overload while providing automatic root cause context when problems arise. The company leverages ClickHouse Cloud for its architecture, opting for a columnar database to manage high cardinality and support real-time metadata correlation, which allows them to track code changes accurately and detect performance degradations. The rearchitected pipeline, now integrated within ClickHouse, enhances scalability and reduces latency by eliminating unnecessary network calls and memory bottlenecks, offering a more streamlined and dependable developer experience. Hud's unique approach bridges business-level issues with engineering details directly within the IDE, making it particularly useful as AI coding agents become more prevalent. As they continue to optimize their system, Hud maintains flexibility with deployment options and focuses on providing engineers with actionable insights into code behavior, aligning with their mission to revolutionize observability.

ClickHouse version 25.10 introduces a variety of enhancements including 20 new features, 30 performance optimizations, and 103 bug fixes. Noteworthy improvements include join enhancements like lazy columns replication, which reduces CPU and memory usage by deferring data duplication until necessary, and the use of Bloom filters to expedite joins by pre-filtering data. The release also adds the capability to push down complex conditions in joins, thus optimizing data retrieval. Automatic column statistics for MergeTree tables enhance query performance by supporting global join reordering, and new SQL updates allow for greater flexibility with operations like IS NOT DISTINCT FROM, negative limits, and offsets. Furthermore, the QBit data type is introduced for vector embeddings, allowing for tunable search precision, and the Arrow Flight integration has been expanded for better server and client compatibility. Finally, late materialization of secondary indices provides options to defer index computation, which can be manually triggered when needed, thereby accelerating data ingestion and processing.

The October edition of What's New in ClickStack introduces several enhancements to improve the observability stack for ClickHouse, including alerting in ClickHouse Cloud, customizable Event Deltas, multiline search support, and new dashboard filters. Alerting now allows real-time alerts across logs, metrics, and traces with seamless integrations and configurable options, while dashboard filters enable easy and interactive exploration without complexity. Event Deltas have been upgraded for greater flexibility, offering advanced analyses for identifying performance regressions. Other improvements include the introduction of Filter Percentages for better data distribution understanding, configurable Kubernetes dashboards for tailored data analysis, and performance optimizations such as request chunking for faster query responses. These updates reflect ClickStack's commitment to providing a robust, intuitive, and efficient observability experience at scale, driven by community contributions and user feedback.

The text discusses the creation of a website designed to allow users to compare various metrics across GitHub repositories, initially intended to encourage managers to address GitHub issues but expanded to a broader analytical tool. It highlights the challenges of using developer productivity metrics, such as commits and pull requests, as reliable indicators of individual work, suggesting they should be secondary metrics. The website, implemented as a single-page HTML application connected to a ClickHouse database, enables users to compare repositories and organizations by typing names, using wildcards, and offers features like horizon charts for dense data visualization. It notes the limitations of directly comparing projects due to differences in issue tracking systems, such as GitHub, Jira, or Phabricator. The website's development, accomplished swiftly with ClickHouse, emphasizes its ease of use for building analytical tools and invites users to try ClickHouse Cloud with promotional credits.

ClickHouse, Inc., renowned for its real-time analytics database, announced its acquisition of LibreChat, an open-source AI chat platform that interfaces with large language models (LLMs), forming the open-source Agentic Data Stack. This acquisition aims to enable organizations to deploy AI agents for analyzing and acting on large datasets using natural language, transforming data interaction into agentic analytics. Companies like Shopify and cBioPortal have already demonstrated the platform's potential, with Shopify using LibreChat to enhance internal AI workflows and cBioPortal leveraging it for advanced cancer genomics research. Fetch, a mobile rewards platform, has also integrated these capabilities to transform consumer data into actionable insights, highlighting the increasing adoption of agentic analytics to democratize data access and accelerate insights. ClickHouse's partnership with LibreChat will enhance its offering with native integration capabilities, extended data visualization, enterprise-grade security, and a tailored semantic layer, positioning the company to lead in the evolving landscape of AI-driven data interaction.

Modal utilizes ClickHouse Cloud to enhance real-time observability for AI workloads across numerous GPUs and containers, overcoming previous scaling issues with data reads and writes. ClickHouse Cloud enables Modal to ingest 1-2 million events per minute and manage around 500 billion logs while maintaining sub-second query speeds, which is crucial for AI infrastructure that supports large-scale GPU workloads for training, inference, and batch processing. Through a seamless Python SDK, developers can deploy workloads without dealing with the complexity behind the scenes. Modal's use of ClickHouse has resulted in the development of several real-time dashboards that offer users detailed insights into function performance, such as execution time and latency trends. These dashboards, powered by a single ClickHouse table, facilitate efficient data querying and provide full lifecycle visibility of function calls. As Modal continues to grow, with event ingestion doubling to 2 million per minute, the team is exploring new features like a billing API and a visual function call graph to further enhance user experience and performance.

ClickStack for ClickHouse Cloud now includes alerting capabilities, achieving feature parity with its open-source counterpart and simplifying setup by eliminating the need for custom configurations. Users in the private preview can create search-based or chart-based alerts directly within the HyperDX UI, covering logs, metrics, and traces. These alerts integrate seamlessly with existing tools like Slack, PagerDuty, and webhooks, allowing teams to respond swiftly to issues such as latency spikes and error rates. The new alerting feature is designed to scale with the performance demands of ClickHouse Cloud and offers flexible options for defining and managing alerts using SQL or Lucene syntax. Looking forward, ClickStack plans to expand its alerting capabilities with broader integrations and advanced logic, including anomaly detection, to support a wider range of use cases.

ClickHouse, Inc., a leader in real-time analytics and AI/ML, has established a Japanese subsidiary, ClickHouse K.K., through a strategic partnership with Japan Cloud, aiming to enhance its presence in Japan's promising AI and data analytics market. This move aligns with Japan's national growth strategy, which emphasizes AI adoption and data infrastructure enhancement as part of its 2025 “Priority Plan for Realizing a Digital Society.” The partnership leverages ClickHouse’s advanced analytics technology and Japan Cloud’s local expertise to support Japanese enterprises in accelerating AI-driven innovation and growth. The collaboration is expected to create new opportunities across various industries, with ClickHouse providing a high-performance, open-source columnar database management system designed for real-time data processing. Trusted by major corporations like Sony and Tesla, ClickHouse aims to drive smarter decisions and insights through its scalable and efficient data platform.

ClickHouse, Inc., a leading provider of real-time analytics data platforms, has established a Japanese subsidiary, ClickHouse株式会社, in collaboration with Japan Cloud to support Japanese companies in leveraging real-time data for swift decision-making and innovation in the AI era. Japan's promising market for AI and real-time data analytics is underscored by government policies emphasizing AI implementation and data infrastructure enhancement. ClickHouse aims to bolster Japanese firms' competitiveness and growth by addressing challenges such as decision-making difficulties, AI model limitations, and operational inefficiencies due to fragmented data systems. ClickHouse's high-performance columnar database management system is designed for large-scale real-time data processing and analysis, enabling rapid insights and decision-making. Globally recognized and adopted by major companies like Microsoft and Sony, ClickHouse has seen significant growth, with its customer base surpassing 2,000 and its annual recurring revenue quadrupling. The partnership with Japan Cloud, known for its successful facilitation of global tech companies' entry into Japan, is expected to further enhance data utilization and corporate growth in the Japanese market.

Constant Contact, a well-known name in email and digital marketing for small and mid-sized businesses, undertook a significant overhaul of its real-time analytics stack due to increasing demands for real-time event processing and advanced reporting capabilities. Initially using Apache Pinot, the company faced challenges with data volume management and maintenance, leading to the decision to transition to ClickHouse in 2024. This transition resulted in substantial performance improvements and cost savings, with ClickHouse offering faster query response times and more efficient data handling due to its use of preads and a log-structured merge tree architecture. Software engineer Abhijeet Kushe highlighted the benefits of ClickHouse, including its ease of setup, efficient data migration, and lower infrastructure requirements compared to Pinot. The migration not only enhanced Constant Contact's technical capabilities but also allowed the team to focus more on developing new features rather than on system maintenance, ultimately strengthening the company's position in the competitive digital marketing landscape.

ClickHouse has acquired LibreChat, an open-source AI chat platform that provides a unified interface for interacting with large language models, to enhance its vision of Agent-Facing Analytics and create an open-source Agentic Data Stack. This integration allows organizations to leverage LibreChat's user-friendly interface and AI framework with ClickHouse's analytical capabilities, facilitating the development of analytics agents capable of processing extensive datasets. Companies like Shopify, cBioPortal, Fetch, SecurityHQ, and Daimler Truck are already utilizing this technology to enhance productivity, democratize AI usage, and enable innovative interactions with data across various applications. The collaboration signifies a commitment to maintaining and expanding the open-source community and infrastructure, while addressing challenges such as LLM inaccuracies by providing accurate contextual data. Future plans include integrating LibreChat within the ClickHouse ecosystem, ensuring seamless user experiences, and evolving the LibreChat Code Interpreter API.

The blog post explores the innovative use of airplane telemetry data, such as position, altitude, speed, and heading, to derive weather information, specifically focusing on wind direction and speed, without directly measuring temperature or wind data. By utilizing the ADS-B Massive Visualizer and storing data in the ClickHouse database, the author demonstrates how comparing various in-flight measurements can yield insights into meteorological conditions. The post delves into the technical aspects of visualizing this data using color spaces like sRGB and OKLCH, where angles are represented by different hues, and discusses how crabbing angles, wind speed, and direction can be calculated. Despite not deriving actual temperatures due to the complexity of formulas, the post highlights the beauty and utility of visualizing derived data, comparing it to real-world weather patterns. The author also shares a method to create video visualizations of these data insights, underscoring the value of real-time, queryable data through ClickHouse and encouraging further exploration with a promotional offer for ClickHouse Cloud.

Event Deltas, a feature in ClickStack's HyperDX interface, is designed to streamline root cause analysis by automatically identifying key attribute differences between fast and slow traces, making it easier to detect performance regressions without relying on labor-intensive manual processes. This tool enhances the open-source observability platform by dynamically comparing trace data properties to highlight changes when performance issues arise, thereby facilitating quicker identification of factors such as deployment versions or specific user segments correlated with slower behavior. Unlike traditional machine learning-based observability approaches, Event Deltas offer an interactive, real-time method for exploring latency anomalies without the need for predefined models or extensive computational resources. By visualizing latency distributions through density heatmaps and using customized metrics for color intensity, Event Deltas help surface significant attributes associated with performance discrepancies. Recent improvements have added flexibility, allowing users to tailor analyses by adjusting visualization parameters, ultimately aiding Site Reliability Engineers (SREs) in efficiently pinpointing and resolving the root causes of system slowdowns.