June 2022 Summaries

ClickHouse users can now quickly install and use Deepnote, a collaborative data notebook for teams to discover and share insights, providing seamless transitions between Python and SQL and first-class support for querying ClickHouse databases directly from the notebook. The integration allows users to explore an interactive example of querying ClickHouse from Deepnote notebooks, with sample queries available, and even participate in a challenge to create something new using the data. This partnership aims to provide a new way of interacting with data, enhancing the value derived from ClickHouse databases.

Building embedded customer-facing analytics can be challenging due to balancing user experience and data architecture. However, using an embedded analytics tool like Cumul.io with a powerful data infrastructure such as ClickHouse can provide a seamless experience for users. ClickHouse offers high-performance and scalability for analytical queries, making it an ideal choice for companies dealing with large amounts of data. With Cumul.io's drag-and-drop UI and API, developers can easily integrate analytics into their platform while retaining control over the feature set. By using ClickHouse as the data infrastructure, companies can improve the speed and performance of their dashboards, providing a better experience for end-users.

Luzmo has partnered with ClickHouse to enable users to build embedded customer-facing analytics. The challenges of building such analytics include balancing user experience and data architecture, while also addressing data security and scalability. Luzmo's embedded analytics platform offers a solution by providing an easy-to-use UI for non-technical users and a powerful API accessible to developers. With ClickHouse as the data infrastructure, performance is improved through its high-speed aggregations and analytics capabilities, making it ideal for customer-facing applications. The combination of Luzmo and ClickHouse enables developers to build empowering customer-facing analytics without dedicating weeks of engineering time, while retaining control over their analytics feature set.

ClickHouse is being used by Superology alongside Kafka to power customer quantitative data, leveraging Google protocol buffers for data serialization. The company's data-informed approach informs business decisions, and they use quantitative data to analyze user behavior and create reports. ClickHouse's built-in Kafka connector and support for protocol buffers enable efficient data ingestion and processing. Superology uses materialized views to filter and transform their data, making it suitable for analysis and reporting. Additionally, ClickHouse facilitates experiments, including AB testing and Bayesian A/B testing, allowing the company to make informed business decisions. The company plans to further integrate ClickHouse with MindsDB to create a machine learning architecture on the database level, enabling more advanced analytics capabilities.

ClickHouse is rapidly developing and adding new features such as semi-structured data support, asynchronous inserts, replicated database engine, RBAC, and others. The development team constantly improves performance and user experience. To compare the performance of different versions, a benchmark was run on an x86_64 AWS m5.8xlarge server with Ubuntu 20.04. The oldest available version used for the benchmark was ClickHouse 1.1.54378 from 2018. The results showed that ClickHouse is 28% faster than the older version. The development team has created a script to run common ClickHouse benchmarks on different hardware, which can be easily extended to add more queries.

ClickHouse has released a new version v22.6, which includes features such as memory overcommit, parallel hash joins, and grouping sets. The release also addresses performance issues with querying incomplete time series data by introducing materialized views that can pre-aggregate recent values while still allowing for real-time queries. This approach combines the benefits of materialized views with the flexibility of querying raw data.