June 2019 Summaries

When running queries in ClickHouse, the profiler often shows the LZ_decompress_fast function near the top, indicating that compression algorithms are a key consideration for query performance. ClickHouse stores data in compressed form to conserve CPU resources, with decompression being a significant portion of the computation time. The optimal compression algorithm can greatly impact query performance, and choosing the right one is crucial for efficient query execution.

ClickHouse has a rich history dating back to 2016 when it was first developed as Yandex.Metrica storage system, which is now widely used by various companies and organizations. The system's evolution over the years has been driven by the need for efficient data processing and analysis in big data applications. In his lecture, Alexey Milovidov delves into the internal implementation of ClickHouse, revealing the design decisions that have made it a popular choice among developers. He shares insights into the architecture and technologies used to power the system, providing a unique look at how Yandex.Metrica storage system has been transformed over time. The event is an opportunity for attendees to learn from Alexey's expertise and gain a deeper understanding of ClickHouse's inner workings.

The 24th global and 3rd Chinese ClickHouse Meetup took place in Beijing on the Dragon Boat Festival weekend, a popular travel event for Chinese people. The event was well-received by hundreds of attendees and thousands more who watched online, with the venue being fully booked as usual. The ClickHouse team used this opportunity to strengthen its bond with the local community in China, delivering an open lecture at the Institute of Computing Technology and engaging in private conversations with prominent corporate users such as ByteDance and JD.com. Notable speakers included William Kwok, CTO of Analysys, Alexey Milovidov from ClickHouse's core developers team, Amos Bird on graph processing, Yan Gangqiang from Golden Data on survey data storage, Percent on a beginner's talk, and Nikolay Kochetov on upcoming query execution pipeline changes. The event also included a pre-meeting with active community members in China.

The 23rd ClickHouse Meetup was held at the CloudFlare office in San Francisco, co-organized by Altinity, with about 35 attendees mostly from the SF and Bay Area. The meetup featured several presentations including a lightning talk on Segment.com's experience with ClickHouse, a discussion on Kubernetes operator for ClickHouse, and an update on new features by Alexey Milovidov, which garnered interest from the audience. A discussion followed on the implementation details of the most appreciated features. The event also provided an opportunity to meet ClickHouse contributors. With increasing demand for such events in the area, Altinity is already planning the next one.

This year's American episode of Percona Live took place in Austin, TX, a waterfront location that welcomed open source database experts amidst hot weather. The ClickHouse community is growing rapidly and has become a popular topic for talks and comparisons. Recent releases have improved its performance, with Alexey Milovidov from Yandex showcasing these enhancements. Altinity provided an entry-level tutorial on ClickHouse, while Ruoxi Sun from PingCAP introduced TiFlash, a column-oriented add-on to TiDB based on the ClickHouse source code. Jervin Real and Francisco Bordenave from Percona discussed moving data around MySQL-compatible storage solutions, highlighting ClickHouse's capabilities in this area. The event also featured talks on modern options for time-series storage, with Evgeny Potapov confirming ClickHouse as a leader in this field. The diverse audience from the US and other countries enjoyed high-quality content, including a tutorial by Alexander Zaitsev and Robert Hodges from Altinity.