NoLambda: A new architecture combining streaming, ad hoc, machine...
summary
Claims a single streaming-first stack (Spark/Scala, Mesos, Akka, Cassandra, Kafka plus FiloDB) replaces Lambda complexity and lowers TCO by letting streaming and ad‑hoc/ML share storage and compute. Demonstrates interactive analytics on the NYC taxi set — 15 million rows scanned on a laptop, heatmaps updating over 650k records in seconds — and a pipeline that writes raw events, continuously retrains models and serves predictions. Production rollout is underway.
AI-written summary · about
chapters (11)
- 0:00Introduction and speaker background
- 3:33Fast data and modern architectures
- 11:46Limitations of Lambda architecture
- 17:54Introducing the NoLambda stack
- 27:06Challenges with Cassandra for analytics
- 28:55Introducing Phb for analytics
- 33:49Phb architecture and benefits
- 37:24Real-world data warehouse with Phb
- 39:33Demo: NYC Taxi data analysis
- 49:04Machine learning with Phb
- 50:38Q&A session
related talks
48mLambda Architecture with Apache Spark
52mLambda Architecture for Real-Time Analytics — Risks and Benefits
23mBuilding Production-Ready Data Pipelines for AI/ML with Lambda Architecture
59mStreaming Big Data w/Spark, Spark Streaming, Kafka, Cassandra & Akka
49mLambda Architecture for High Data Volume and Fast Business Response
01h 12mImplementing the Lambda Architecture with Spring XD
04mInterview
35mStreaming and Real-time Analytics