US-based 2020 ClickHouse meetups started the year with a bang: two events in San Francisco and New York City. Here is a summary of the talks including links to presentation slides.
San Francisco ClickHouse Meetup – 5 February
CloudFlare sponsored the February meetup at their offices on 101 Townsend Street in San Francisco. It’s a great venue with plenty of space for the 45 or so people who were in attendance. We had a total of four talks starting with two lightning talks from CloudFlare engineers.
Query profiling without EXPLAIN — Arman Baratifar described how to gather useful query profiling information from ClickHouse. ClickHouse does not have an EXPLAIN command, but between summary stats and the query log you can figure out what your query is doing and make it faster.
ClickHouse Mark Cache: 5 Minute Primer — Mik Kocikowski gave a quick overview of how ClickHouse uses marks to locate data as well as tips to ensure marks are cached to speed up query performance.
What the heck is ClickHouse tiered storage? — Robert Hodges of Altinity introduced tiered storage, a new feature of ClickHouse that allows users to put hot data on one type of storage, then automatically move it to another type as the data ages. The talk explains the different policies as well as our dev roadmap.
The Most Fresh ClickHouse Features — Alexey Milovidov of Yandex presented a roundup of recent additions to ClickHouse plus coming attractions. The fresh features include tiered storage, DDL definition of dictionaries, parallel parsing of data formats, partial merge joins, DateTime64 types and many more. Coming attractions include RBAC and storage/compute decoupling. ClickHouse is continuing to improve quickly!
We would like to offer thanks to CloudFlare, who have been great sponsors for multiple ClickHouse meetups, and especially Mik Kocikowski for helping organize things at the venue. Also, to the Yandex ClickHouse team including Alexey Milovidov, Ivan Blinkov, and Nikolai Kochetai who traveled to California to give presentations.
Our next meetup will be in April and we’re looking for a venue. Maybe the South Bay? Send a message to the organizers or firstname.lastname@example.org if you want to help out or present.
New York City ClickHouse Meetup – 11 February
Uber Engineering sponsored the first ever NYC ClickHouse meetup at their offices located at 1400 Broadway in Manhattan. We had around 50 attendees and a bunch of great talks.
ClickHouse Introduction — Alexey Milovidov of Yandex gave an overview of ClickHouse including significant features, development history, and an explanation of the eternal question for new users: why is ClickHouse so crazy fast? It was a great setup for the next talk.
Fast, Scalable and Reliable Logging at Uber with ClickHouse — Chao Wang of Uber described a new, ClickHouse-based log analysis system that the Uber observability team is deploying. It included an overview of the logging problem, why the team chose ClickHouse (5x faster than Elastic, among other reasons), and the logging system architecture including key ClickHouse features that make possible. Uber operates at enormous scale, so this presentation will be interesting to anyone doing logging.
Lightning talks. Between presentations, Robert Hodges did a lightning talk on tiered storage using TTL policies (see above), and Alan Braithwaite described Segment’s use of ClickHouse in their systems.
ClickHouse for Machine Learning — Nikolai Kochetov of Yandex walked through support for machine learning in ClickHouse. ClickHouse supports linear regression, logistic regression, and catboost, all of which are accessible from SQL. Nikolai showed how to train models and run inference.
As noted above this was the first-ever ClickHouse meetup, and we hope the start of a great tradition. We would like to thank Mark Bonano and Chao Wang for the leg work to get it set up and to Uber for providing such a great venue. We plan to work with Uber, Segment, and the many others who were present to organize future meetups in the New York area. If you are interested in either presenting or sponsoring the next meetup, please send an email to email@example.com. See you soon!