Spark Analytics on Cassandra Data
summary
Explains why Cassandra’s layout — partition key for node placement and ordered clustering column inside partitions — favors fast single-row and recent-N queries but frustrates ad‑hoc aggregation; recommends Spark + the Spark‑Cassandra connector (used inside the Spark shell and DataStax Enterprise) to run distributed analytics. Live demo joins a CSV to a Cassandra table, builds pair RDDs, uses reduceByKey and mapValues to compute per‑movie averages (example result ≈ 8.53), showing the pipeline runs and feels like ordinary code. Performance notes: Cassandra aims for single‑digit ms ops; Spark is batch, often faster than Hadoop, and not strictly in‑memory.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 3:40Why Cassandra
- 6:22Data Model
- 10:25Data Structure
- 17:18Need for Spark
- 21:28Bad things about Spark
- 22:45RDDs
- 24:05Cluster Manager
- 27:56Schema
- 29:30Distributed Database
- 30:44Spark Shell
related talks
52mSolving Analytical Problems using Apache Spark
38mReal-time analytics with Apache Cassandra & Apache Spark
54mHarnessing the Power of Spark & Cassandra within your Spring App
55mSolving Classical Data Analytic Task by Using Modern Distributed Databases
21mLightning fast cluster computing with Cassandra and Spark
31mApache Cassandra Analytics: A Recipe to Move Petabytes of Data
33mTo Spark or Not to Spark
05mInterview