Apache Cassandra Analytics: A Recipe to Move Petabytes of Data
summary
Two open source components enable bulk moves of petabytes by streaming SSTables instead of row I O the Cassandra Analytics Library runs on Spark to deserialize SSTables locally while the Cassandra Sidecar streams files and triggers JMX import on nodes the pipeline skips Cassandra s read/write path, relies on sendfile, enforces token metadata consistency, includes throttling, supports Cassandra 4 and 4.1 with plans for 5.0, benchmarks show about 30× faster, code is on GitHub but not yet a 1.0 release
AI-written summary · about
chapters (9)
- 0:00Cassandra analytics intro
- 1:07Problem statement today
- 3:18Bulk reads/writes cost
- 9:17Analytics architecture
- 13:15Bulk read workflow
- 15:25Bulk write workflow
- 17:34Getting started
- 20:22Resources & community
- 21:04Q&A
related talks
40mSpark Analytics on Cassandra Data
55mSolving Classical Data Analytic Task by Using Modern Distributed Databases
19mBig Data
18mOur Analytical Data Journey – from Paper to Petabytes
35mReplacing Cassandra: A Digital Transformation for the World’s Largest Digital Identity Network
19mZero to Scale: Telemetry pipeline with Apache Cassandra
53mApache Cassandra. Inception - all you need to know
33mAnalytics in the age of the Internet of Things #bbuzz