neds.gif (1190 bytes)

New England Database Society

Friday, December 9, 2005

sponsored by Sun Microsystems

sunlogo.gif (4979 bytes)

NEDS

Database Systems and Storage Virtualization

Ken Salem
University of Waterloo

Friday, October 21, 2005, 4:00 PM
Volen 101, Brandeis University

(preceded by a wine and cheese reception at 3:00 pm)

Abstract:

In this talk, I will first present some recent work on managing two-tier I/O buffer caches.  Modern storage servers typically have large memories in which they cache data blocks.  Storage clients, such as database management systems, also rely heavily on caching.  This creates a two-tier cache.  The storage server's cache is difficult to manage effectively because of interactions with the first-tier cache in the database system.  I will describe a technique that can improve its effectiveness.  The technique relies on "write hints", with which the storage client (e.g., the database system) provides the storage server with an indication of its reason for sending data to the server.  The server can then exploit these hints to more effectively manage its cache.

In the second part of the talk, I will argue that the two-tier caching problem is one manifestation of a larger issue.  Database systems are losing control of the physical environment in which they are expected to operate.  For example, instead of dedicated, locally-attached disks, database systems are expected to use shared, virtualized, and externally managed storage.  The same thing is happening to the rest of the physical environment in which the database system runs.  This has many implications.  I will illustrate some of the issues using one or two examples focusing on the interaction between database systems and externally-managed, virtualized storage.

Speaker Bio:

Ken Salem is an associate professor in the School of Computer Science at the University of Waterloo, where he has been since 1994. He received his B.Sc. in electrical engineering from Carnegie Mellon University, and his Ph.D. in computer science from Princeton University in 1989.  His primary research interest is data management systems.


Maintained by Dina Goldin dqg AT cse.uconn.edu