A New Way to Restore ClickHouse After ZooKeeper Metadata Is Lost

A New Way to Restore ClickHouse After ZooKeeper Metadata Is Lost

Every ClickHouse user has experienced data loss due to user error or operational failure with Zookeeper at least once in their lifetime. If a failure resulted in metadata loss, ClickHouse tables can turn into pumpkins–you can query but you cannot insert more data. But as of ClickHouse version 21.7, this has changed with a new SQL Command. Keep reading to learn more.

Webinar Slides. MORE secrets of ClickHouse Query Performance

By Robert Hodges and Alexander Zaitsev
Slides for the Webinar, presented on May 27, 2020
Webinar recording is available here

ClickHouse is famously fast, but a small amount of extra work makes it much faster. Join us for the latest version of our popular talk on single-node ClickHouse performance. We start by examining the system log to see what ClickHouse queries are doing. Then we introduce standard tricks to increase speed: adding CPUs, reducing I/O with filters, restructuring joins, adding indexes, and using materialized views, plus many more. In each case we show how to measure the results of your work. There will as usual be time for questions as well at the end. Sign up now to polish your ClickHouse performance skills!

Strength in Numbers. Introduction to ClickHouse Cluster Performance

Strength in Numbers. Introduction to ClickHouse Cluster Performance

ClickHouse clusters apply the power of dozens or even hundreds of nodes to vast datasets. In this webinar, we’ll show you how to use the basic tools of replication and sharding to create high performance ClickHouse clusters. We’ll study the plumbing of inserts into sharded datasets and how to determine the correct number of shards for your desired writes. We’ll similarly look at distributed queries and show how to scale read capacity to desired levels using replicas. Finally, we’ll look at techniques for scaling up both shards and replicas to accommodate growth in your dataset.

Webinar Slides. Strength in Numbers: Introduction to ClickHouse Cluster Performance

By Robert Hodges and Alexander Zaitsev
Slides for the Webinar, presented on Apr 29, 2020
Webinar recording is available here

ClickHouse clusters apply the power of dozens or even hundreds of nodes to vast datasets. In this webinar we’ll show you how to use the basic tools of replication and sharding to create high performance ClickHouse clusters. We’ll study the plumbing of inserts into sharded datasets and how to determine the correct number of shards for your desired writes. We’ll similarly look at distributed queries and show how to scale read capacity to desired levels using replicas. Finally, we’ll look at techniques for scaling up both shards and replicas to accommodate growth in your dataset.