
DB2 for Linux, UNIX, Window by Tim Vincent, IBM Fellow, Chief Architect for DB2 for LUW. Most of this talk was unfortunately confidential. The only section not under NDA was about DB2 pureScale (DB2 for LUW's new OLTP shared disk clustering solution) that allows unlimited scalability. This figure shows almost linear scalability, at least up to 128 nodes. Initially it is available only for AIX servers, but others DB2 servers will be available in the future.

eXtreme Analytics Platform by Hamid Pirahesh, IBM Fellow, Senior Manager, Exploratory Database at IBM Research. Hamid delivers the best presentation in these Gold briefings. See my blog from last year about his presentation style and innovative content.
Hadoop evolution as an open-source data engine for BIG applications for Yahoo!, Facebook, Amazon, and AOL. It is using thousands of servers using petabytes of data. Huge appetite for extreme analytics for many large corporations. Need for a eXtreme Analytics Platform (XAP) that takes wide spectrum of unstructured, structured, and semi-structured data from diverse external sources into a MapReduce Hadoop++ data engine. IBM's motivation is to provide corporate IT with a value-enhanced alternative to each user doing their own thing on their own personal data cache from often semi structured and unstructured external sources. To support a Java Analytics Apps Server, IBM has proposed JAQL as an open query language for JSON data model on XAP. To perform for the data analysis, the R statistics language, along with SPSS, will be used. Hamid did a demo of running analytics against CO2 observations (simulating real data from XAP customer engagements) using both R and SPSS. He then explained a case of exposing unusual financial activities for the SEC.
Posted October 24, 2009 1:50 PM
Permalink | No Comments |
Leave a comment