Next-Generation Log Analysis with ClickHouse

Come learn about ClickHouse (https://github.com/ClickHouse/ClickHouse), an open source high-performance column-oriented database management system for analytics, at a meetup on February 11, 2020 at the Uber office, Broadway 1400, NYC on the 12th floor.

ClickHouse can generate custom data reports in real time and process billions of rows and dozens of gigabytes of data per single server per second. It works up to a thousand times faster than traditional approaches. ClickHouse is linearly scalable, hardware-efficient, fault-tolerant, and can be deployed across multiple data centers. Among other features, ClickHouse offers a user-friendly SQL query dialect with a number of built-in analytics capabilities. Through interactive talks, attendees will learn about product features, how ClickHouse can benefit them, and how to use this system in practice. Attending the ClickHouse meetup is free.

Talk 1: ClickHouse Introduction

Alexey Milovidov is a software engineer at Yandex and leads the ClickHouse development team. Before joining the ClickHouse team, Alexey led the Metrica engine development team at Yandex and he has been an engineer at Yandex for almost 12 years.

Abstract

During this talk, we will discuss ClickHouse’s development history, main use cases, and explain how ClickHouse works to provide sub-second query latencies even on petabyte-sized data sets.

Talk 2: Fast, Scalable and Reliable Logging at Uber with Clickhouse

Chao Wang is a software engineer at Uber and leads the log analysis platform team. Chao is also a core contributor to M3, Uber’s open source metrics platform, and has worked in Observability for the last four years.

Abstract

The log analysis platform at Uber ingests and serves petabytes of data. With the need to support streaming data analysis, ClickHouse is a key underlying technology to support Uber’s performance and scaling requirements. In this talk, we will discuss how ClickHouse has been incorporated into the log analysis platform and the benefits it has brought to Uber’s larger log analysis initiative.

Talk 3: ClickHouse for Machine Learning

Nikolay Kochetov has been a ClickHouse core developer for his entire tenure at Yandex. Prior to joining Yandex, Nikolay studied at MIPT university.

Abstract

In this talk, we will cover how ClickHouse can be used to train simple models and run inference on them. We will also demonstrate how we’ve integrated ClickHouse with the CatBoost gradient boosting library.