Streaming analytics is a valuable capability that helps organizations extract insights from log data in real-time. Building a modern streaming analytics architecture on AWS consists of five logical layers: source layer, ingestion layer, storage layer, processing layer, and destination layer. Amazon S3 can be used as a data lake for storing raw streaming data, providing high availability, unlimited scalability, and low data storage costs. Implementing data indexing and compression inside the data lake accelerates query performance, reduces costs, and overcomes data retention limitations. Additionally, using Amazon S3 Express One Zone provides low-latency applications with fast request latency and lower compute costs. Finally, enabling multi-model analytics on streaming data in S3 with Chaos LakeDB allows for true multi-model data access with support for search analytics, SQL querying, and GenAI analytics. These best practices can help optimize a streaming analytics architecture, reduce costs, and extract powerful insights from data.