|
Panel 1: Database Kernel Research: What,
if anything, is left to do?
Tuesday March
30, 4:00 – 5:30 pm, Alcott room
Chair: Dave Lomet (Microsoft Research)
Panelists: Michael Brodie (Verizon Corporation), David
DeWitt (University of Wisconsin), H.V.Jagadish (University of
Michigan), and Gerhard Weikum (Max Planck Institute)
By "kernel" we mean access methods,
buffer management, concurrency control, recovery, disk management, etc. There are many facets of the database
landscape that might impact the utility of database kernel research:
·
changing hardware: cheap RAM, cheap disks in vast sizes; greater
relative latency between processor and main memory; stable RAM, e.g. MRAM, or
simply battery backup;
·
changing configurations: clusters, scalability, partitioning, data
sharing vs. shared-nothing, cyber bricks; disk mirrors & RAID; storage
controllers; NAS; wide scale distribution, e.g. the web;
·
changing requirements: more(ever higher) availability (faster failover,
more redundancy) and scalability; more security; "trustworthy
computing"; privacy;
·
new forms of data: XML data; temporal data; object/relational;
object oriented
·
new forms of data
organization: horizontal vs. vertical
partitioning; master-detail clustering; materialized views; broken mirrors;
partial views; automatic reorganization;
·
new forms of indexing: multi-attribute; hashing; nearest neighbor;
temporal;
·
new applications to
support: with new concurrency control
and recovery; new transaction models; queues in database; main-memory databases
with integrated apps.
We hope that, in addition to being entertaining,
the panel would produce an interesting research agenda for database kernels.
Panel 2: Querying the Past, the Present, and the Future
Wednesday March 31, 2:00 – 3:30 pm, Alcott room
Chair: Dieter Gawlick (Oracle Corporation)
Panelists: Adam Bosworth (BEA), Michael J. Franklin (University
of California, Berkeley), Christian S. Jensen (Aalborg University,
Denmark)
Existing database
technology is focused on managing the current state of data and on sharing
data on request only. This focus fails to reflect current and emerging
business needs. Today’s businesses need access to the complete history of
data (“we need to know what was known and/or what happened when”). Businesses
also need the ability to distribute relevant information ASAP to the right
group of people, or even to react automatically in real time (“we need to be
on top of things”). Frequently cited examples include ePC/RFID (electronic
Product Code/Radio Frequency Identifier), RTE (Real Time Enterprise), and BAM
(Business Activity Monitoring).
Two fundamental
enhancements are needed: The ability to automatically create and provide
access to the complete history of data, as well as the ability to inform
interested parties of relevant changes as soon as they happen and/or to activate
real-time agents. This panel will focus on necessary enhancements of database
technology with respect to data model, query language, and operational
characteristics, and explore how a combination of ideas from a variety of
existing disciplines can help in meeting these new challenges. The list of
relevant technologies includes temporal database technology, streams SQL,
event management, publish/subscribe technology, agent management, and
information distribution. We expect that some suggestions will be in conflict
with current, well-accepted approaches.
Panel 3: Database
Research in the Current Millennium
Thursday April 1, 4:00 – 5:30 pm, Alcott room
Chair: Daniela Florescu (BEA Systems)
Panelists: Ioana Manolescu (INRIA, France), Anastassia Ailamaki
(Carnegie Mellon University), Jai Shanmugasundaram (Cornell
University), Zack Ives (University of Pennsylvania)
(all panelists have
Doctorates awarded since 1/1/2000)
The database world today (or
better—the information world) is totally different from the peaceful days when
the database research field was created. Moreover, it is in constant
movement. Lets list some of the changing factors. First the Internet forever
changed our lives. Then came XML as an innocent character-by-character
UNICODE syntax, and that changed all the rules. Then Web Services arrived,
invented by marketing departments in the middle of the boom, and only later
taken seriously by vendor capitals and technologists. Now mobile computing
and messaging are pervasive. And, finally, we see a shift in perspective due to
the dramatic reduction of hardware costs.
In this new world,
information is not just in databases. Information is mobile, flexible,
mirrored in a variety of logical and physical forms, moving from platform to
platform, evolving, being copied, modified and replicated, and later
reintegrated. Information is alive and active. Web services are awake and
listening, automatically capturing data, sending and receiving data to and from one another, triggering
automatic actions. Databases as we know them are only a resting place for data
at the end of a long journey.
So where are we in this
changing world? And where are we going?
Panel 4: Outrageons
Ideas and/or Thoughts While Shaving (Plenary Panel)
Friday, April 2, 11-12:30, Alcott room
Moderator:
Mike Stonebraker
For this closing panel discussion, we will
recruit a collection of participants from the attendees and organizers. Each will agree to present one or more
outrageous ideas that are too wacky to get funded and/or incapable of being
turned into least-publishable units (LPUs).
Less adventuresome panelists can present their pet peeve about
research activities pursued by other in the DBMS community. Risk averse
panelists can discuss more mundane problems which they would like to work on
if they had more time or were excused from department committees.
Each panelist will be
encouraged to comment on the other panelist’s ideas. At the conclusion of the discussion, we
will have an award for the most outrageous idea as well as the most
thoughtful one.
|
