JALURI 17,453 SUMMARIES / 50 SOURCES
SEARCH LAST PASS 07:00 ATOM
ALL/ SOFTWARE DEVELOPMENT

Netflix Tudum Architecture: from CQRS with Kafka to CQRS with RAW Hollow

netflixtechblog.com · · RSS

Netflix's Tudum platform transitioned from a CQRS architecture with Kafka to using RAW Hollow, enhancing data propagation speed, reducing latency, and improving user experience by leveraging in-memory caching and strong read-after-write consistency.

MAIN POINTS
  1. Tudum.com is Netflix’s fan destination offering exclusive content and interactive experiences.
  2. Initial architecture used CQRS with Kafka, causing delays in content updates due to eventual consistency.
  3. RAW Hollow, an in-memory database, improved data access speed and reduced architecture complexity.
  4. Migration to RAW Hollow decreased homepage construction time from ~1.4 seconds to ~0.4 seconds.
TAKEAWAYS
  1. CQRS can scale well but may introduce delays due to eventual consistency.
  2. In-memory caching with RAW Hollow significantly reduces I/O and improves performance.
  3. Eliminating sequential operations can greatly enhance response times.
  4. RAW Hollow allows for efficient dataset management with strong read-after-write consistency.
READ THE ORIGINAL