Lecture: Cassandra on Steroids
Migrating from memcached to Cassandra
Hear a war story how trivago migrated a high-throughput Cache from memcached to Apache Cassandra with intermediate success, fulminating failures, surprising Cassandra bugs and a final victory doing 1 million writes per second.
Apache Cassandra is a widespread distributed NoSQL database. It is well known for its high throughput, but in 2012 researchers from the University of Toronto concluded that "this comes at the price of high write and read latencies."
Is this (still) true in all scenarios, or can Cassandra deliver low latency? This talk will walk you through the Cassandra Read- and Write-Path, Bloom-Filters, tuning compaction (TWCS), optimizing configuration and the influence of client libraries on performance.
Hear a war story how trivago migrated a high-throughput Cache from memcached to Apache Cassandra, with intermediate success, total failures, surprising Cassandra bugs and a final victory doing 1 million writes per second, that are also replicated. How many Cassandra instances are required to make 1 Million writes per second happen?