Author: Stephane Combaudon.
Using full index scans are not always faster than a full table scan. Stephane explores the situations in which a sequential read through of a full table performs better than using randomly-read full index.
Author: Matt O’Keefe.
150 TB of log files. Per day. eBay has to deal with an enormous amount of data that is traditionally unformatted and tricky to deal with en masse. They have needed to formalize all logging and created a system using OLAP, Hadoop and H base to deal with all levels of logs. The final result is combined into a massive cube for data mining.
In the bid for success in online marketshare, information is very important. Monitis has published results from their user survey, providing hard numbers to the understood fact that customers want responsive websites. With other questions in the survey this is a useful resource for e-commerce providers.
Author: Jon Vlachogiannis.
If the existing BigData solutions don’t seem a best fit to your situation, you may have considered rolling your own solution. BugSense did this with their in-memory data, writing a LISP language within C to avoid VM use, using erlang for inter-node communications, and modifying TCP_TIMEWAIT_LEN in the kernel.
Author: Paul Robert Lloyd.