Database Research.

What Next?

Phil BernsteinMicrosoft Research

May 29, 2008

© 2008 Microsoft Corporation

Database Engines ReduxDBMS architecture hasn’t change much since System R and Ingres.

Locking, logging, B-trees, ….

DB engines aren’t scaling with h/w advancesMulti-core, huge RAM, flash, disk capacity, networks

We need to reconsider all assumptions.Full functionality (txns, SQL) for TPCMust try out new designs

Goal: Better cost/perf and scaleoutEspecially for large clusters

2

Explore Holes in theIntegration Solution Space

Approx

Precise

Text

Format

ted

GIS…

Process

TypeP

recision

Time

BoxedIncr

emental

ly

Develop

ed Carefully

Enginee

redFully

Automat

ic

3

Apps O

ntlgies

DB Runtime Should Map All Actions and Constraints

QueriesUpdatesPeer-to-peerProvenanceAccess Control

Integrity constraintsSynch logicBusiness logicDebugging

ErrorsIndexingNotificationsBatch loadingData exchange

Customer

Order

ScheduledDelivery

Product

Salesperson

Mapping

Actions & Constraints Actions & Constraints

Interpret or CompileTarget Source

4

ExampleMap target constraints to source constraints

Class0

Class1 Class2 Relation1 Relation2

Target Source

(Class1 as Class0) (Class2 as Class0) = ?

5

More Interdisciplinary Search Research

Information RetrievalDatabase SystemsInformation ExtractionNatural Language ProcessingInference enginesHCIWisdom of crowds

6

Green Database SystemsPower networks are built for peak load

Reduce the peak energy demand of a DBMS

More background processing to avoid on-line work

More materialized viewsDifferential files to avoid eager view refresh

7

8