Scaling and Partitioning High Performance Transactional Databases for The Cloud

Date: 
Friday, February 25, 2011 - 8:46pm

UCSB COMPUTER SCIENCE DEPARTMENT PRESENTS:

Wednesday, March 9, 2011
11:00 AM – 12:00 PM
Computer Science Conference Room, Harold Frank Hall Rm. 1132

HOST: Divy Agrawal

SPEAKER: Sam Madden
MIT, Computer Science and Artificial Intelligence Laboratory

Title: Scaling and Partitioning High Performance Transactional Databases for The Cloud

Abstract:

In this talk I will summarize recent work we have done at MIT on building high performance transaction processing systems as a part of a cloud-based database service. Such a service will need to host a number of tenant database systems, and allocate resources to each to allow it to meet customer’s transaction throughput and latency goals. I will describe two key technologies meant to address the challenges that arise in such a system. First, I will describe Kairos, a system for profiling the disk and RAM utilization of a database tenant and allocating it to one of several servers in a database cluster based on its compatibility with other tenants over time. Second, I will describe Schism, a technique for partitioning the load of one OLTP database into a number of partitions such that each partition receives about equal load. This partitioning facilitates fine-grained placement with Kairos, allows scaling of the database service to very large tenants, and enables fast migration of load from one server to another by moving partitions that are much smaller than an entire database.

Bio:

Samuel Madden is an associate professor of electrical engineering and computer science in MIT’s Computer Science and Artificial Intelligence Laboratory. His research interests include databases, sensor networks, and mobile computing. Research projects include the C-Store column-oriented database system, and the CarTel mobile sensor network system. He received his Ph.D. from the University of California at Berkeley in 2003 where he worked on the TinyDB system for data collection from sensor networks. Madden was named one of Technology Review’s Top 35 Under 35 in 2005, and is the recipient of several awards, including an NSF CAREER Award in 2004, a Sloan Foundation Fellowship in 2007, best paper awards in VLDB 2004 and 2007, and a best paper award in MobiCom 2006.