DB system design for new hardware and sciences Anastasia Ailamaki École Polytechnique Fédérale de Lausanne and Carnegie Mellon University
introspective
parallelism sharing Uni-processorMulti-core Multi-processorCluster Exploit max parallelism and sharing simultaneously
Moore’s Law = cores Performance burden shifts to software UltraSparc T2 Power CRS-1 (Tensilica)
Multi-core challenges for DBMS CMP-aware parallelism in OLTP –Efficient synchronization –Highly concurrent algorithms CMP-aware sharing in BI –Eliminate redundancy with work sharing –Improve locality in query operators … But programmers are not multithreaded
sciences
Challenges: Complexity AND size Alliez et al, INRIA, SIGGRAPH’05 Brain Mind Institute, EPFL Automate DB Design Computational Support Understanding Data
Summary Challenge #1: exploit hardware –Parallelism, sharing maximized simultaneously –Infrastructure to parallel thinking&programming Challenge #2: serve sciences –Reduce complexity through abstraction –Manage large datasets on large computers
“Multicore: This is the one which will have the biggest impact on us. We have never had a problem to solve like this. A breakthrough is needed in how applications are done on multicore devices.” – Bill Gates “It’s time we rethought some of the basics of computing. It’s scary and lots of fun at the same time.” – Burton Smith