Brian Bulkowski - BigDataNYC 2014 - theCUBE - #BigDataNYC
summary
Aerospike unveils a Hadoop connector (input/output format above HDFS) that lets Hadoop jobs emit results straight into Aerospike or query live records so batch analytics become immediate front‑end inputs. Advertising use case: AppNexus switches to Aerospike, replaces many Cassandra nodes, runs auctions with ~100 bidders and performs sub‑10ms auction pricing for each impression. Claim: Aerospike is roughly 10x faster than Cassandra and powers millions of ad decisions per second; source code and examples are on the company site.
AI-written summary · about
chapters (18)
- 0:00Introduction to BigDataNYC
- 1:02Oracle vs Hadoop World comparison
- 2:05Aerospike’s Hadoop integration
- 3:07Real-time pricing in retail
- 5:16Enterprise pressure from internet giants
- 6:59Retail innovation and competition
- 8:21Real-time personalization in stores
- 9:03Aerospike’s role in real-time apps
- 10:47Defining real-time performance
- 11:51Hadoop connector for Aerospike
- 12:54Google-inspired innovations
- 13:56Advertising trading system case study
- 15:38Scaling challenges and solutions
- 16:41Silicon Valley market bubble concerns
- 18:27Mobile market consolidation
- 19:30Future of mobile and big data
- 20:33Aerospike’s growth and Go language
- 21:57Closing remarks
