We will look at the evolution of this design decision and how Netflix went about implementing it.

ByteByteGo provides tutorials, articles, and resources for learning and mastering the Go programming language, covering topics such as syntax, concurrency, and best practices. Developers can learn about Go programming fundamentals, web development with Go, and building scalable applications using Go's powerful features and standard library.

ByteByteGo

Netflix redesigned their Tudum platform architecture to support 20 million users by replacing a traditional CQRS implementation with RAW Hollow, an in-memory object store. The original design used Kafka and Cassandra with caching layers, causing delays in editorial previews due to eventual consistency. By embedding RAW Hollow directly into microservices, they eliminated external datastores and reduced page construction time from 1.4 seconds to 0.4 seconds while enabling near-instant editorial previews. The compressed in-memory approach stores three years of data in just 130MB while maintaining strong consistency options for critical workflows.

How Netflix Tudum Supports 20 Million Users With CQRS

Build MCP servers that perform with Gram by Speakeasy (Sponsored)

<p>For those reading this feeling the need to google a bunch of terms…</p>
<ul>
<li>CQRS is the idea that you use different models for updating your data and displaying your data. ( Exceptional use case would be a fully transactional system like banking software).</li>
<li>RAW Hollow is a noSql/Object storage that is local to the app and extremely fast.</li>
<li>Apache Kafka is a log-based message broker (streams of queued messages on disk provide reliability and replayability)</li>
<li>Apache Cassandra is a distributed NoSQL database for datasets that are highly available and grow quickly, but does have some cons. ( <a href="https://www.geeksforgeeks.org/system-design/system-design-when-to-and-when-not-to-use-cassandra/" target="_blank" rel="noopener nofollow">https://www.geeksforgeeks.org/system-design/system-design-when-to-and-when-not-to-use-cassandra/</a> )</li>
</ul>
<p>RAW Hollow excerpt:
“By leveraging compression, despite being in-memory, RAW
Hollow allows scaling datasets up to a 100 million records per
entity. Reads are eventually consistent by default, and completely
local to the application, ensuring extremely low latencies and high
availability. RAW Hollow allows applications to query millions
of records in milliseconds” – source: <a href="https://hollow.how/raw-hollow-sigmod.pdf" target="_blank" rel="noopener nofollow">https://hollow.how/raw-hollow-sigmod.pdf</a></p>
