Pivot
summary
Amazon S3 stores 566 billion objects and an internal catalog of 1.5 billion items that receives about 50 million updates a week. Teams spin up EMR clusters frequently — one team every 30 minutes — to reclassify catalog items in under an hour instead of days or weeks, at much lower cost. Yelp runs 100-node EMR jobs from S3 for spelling and analytics (80 engineers spin up ~50–250 clusters per week). Cycle Computing rents 30,000 cores on demand for ~$1,300/hour instead of a projected $18M build.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:27Amazon Big Data
- 1:49Data Oriented Problems
- 2:48Traditional Big Data
- 4:34Big Data
- 5:07AWS
- 6:45Cycle Computing
- 7:10Conclusion
related talks
22mTurn Data Into Insights
34mBest Practices for Implementing a Data Lake in Amazon S3
21mOptimize and manage data on Amazon S3
05mAWS On Air ft. Amazon EMR Serverless Quick Tip 2
41mEMR Serverless
28mInnovations in AWS analytics: Data processing
26mPetabyte Scale Datalake Table Management with Ray, Arrow, Parquet, and S3
22mHow Rocket Companies modernized their data strategy on AWS