Separating Hadoop Myths from Reality

on YouTube ↗10m

" summary

Claims that commercial Hadoop distributions all behave the same are rebutted with architecture: MapR removes Java layers and the HDFS single-write choke, producing consistent low latency versus the spike-prone HDFS setup shown. Production numbers: Solutionary analyzes one trillion log lines, Rubicon processes 90 billion ad auctions per day, retailers run 2,000+ node clusters, finance firms 1,000+ nodes. Use cases include fleet optimization for waste management, facial-recognition vending, Climate Corporation weather analytics, and India’s Aadhaar storing petabytes of biometrics. Price and availability differences are quantified.

AI-written summary · about

" chapters (5)

1. 0:00Start
2. 0:28Hadoop Distributions
3. 3:19Architecture Comparison
4. 3:34Architecture Results
5. 4:41Production Success with Hadoop

" related talks

• 57mBig Data: Threat or Opportunity?Shaun Connolly · PASS Business Analytics Conference 2013
• 47mSolving Real & Big (Data) Problems Using HadoopEva Andreasson · GOTO Chicago 2014
• 47mHadoop: Lessons LearnedTorsten Curdt · Berlin Buzzwords 2012
• 29mBig Data AnalysisSerhiy Masyutin · JavaDay Lviv 2015
• 07mindustries needing data disruptionDavid Richards · Strata + Hadoop World London 2015
• 05mHadoop in 5 Minutes or LessJohn Schroeder · Strata Conference 2014
• 56mResistance is Futile: What You Need to Know about Big DataMichelle Ufford · PASS Business Analytics Conference 2015
• 19mDemystifying Big DataMohit Saxena · Great Indian Developer Summit 2015