This document provides an overview of Weather.com's analytics architecture using Apache Cassandra and Spark. It summarizes Weather.com's initial attempts using Cassandra, lessons learned, and its improved architecture. The improved architecture uses Cassandra for streaming event data with time-window compaction, stores all other data in Amazon S3 for batch processing in Spark, and replaces Kafka with Amazon SQS for event ingestion. It discusses best practices for data modeling in Cassandra including partitioning, secondary indexes, and avoiding wide rows and nulls. The document also highlights how Weather.com uses Apache Zeppelin notebooks for data exploration and visualization.