You are on page 1of 39

Sem-ODB: Semantic Object DBMS

FIU High Performance Database Research Center Dr. Naphtali Rishe, Director
1

SEMANTIC DATABASE PRESENTATION OUTLINE

Introduction DEFINITION Features Benefits Market Demonstration Summary


2

SEMANTIC DATABASE DEFINITION

THE FLEXIBLE DATABASE MANAGEMENT SYSTEM THAT STORES THE MEANING OF INFORMATION AS FACTS ABOUT OBJECTS.

SEMANTIC DATABASE DEFINITION


Semantic Binary Model Object-Oriented Features Semantically-Enhanced Object-Relational A Collection of Facts Arbitrary Relationships Storing the Inherent Meaning of Information Information in its Natural Form Information Handling System
4

SEMANTIC DATABASE DEFINITION


SEM-ODB
Meaning of Information is Stored Relationship Between Classes No Restriction on Data Type/Size Any Query Can be Run Ad Hoc Any Relation Can be Viewed No Keys are Required

RDBMS
Meaning of Information is Lost Relationships not Supported Restricted Data Type and Sizes Most Queries Have to be Predicted Expensive Joins are Needed Keys are Required and are Static

SEMANTIC DATABASE PRESENTATION OUTLINE

Introduction Definition FEATURES Benefits Market Demonstration Summary


6

SEMANTIC DATABASE FEATURES

Semantic Database Design Optimal Processing Algorithms Efficient Storage Techniques Application Schema Design Methodology ODBC/SQL Compliance Semantic SQL Internet/WEB Enabled
7

SEMANTIC DATABASE FEATURES


Exceptional usability and flexibility Shorter application design and programming cycle Provides user control via an intuitive structure of information Empowers end-users to pose complex ad hoc decision support queries Superior efficiency-Highest level of optimization Massive reduction in storage size for large applications, such as Data Warehouses Directly supports conceptual data model of the enterprise Internet-integrated

SEMANTIC DATABASE FEATURES


Semantic view mirrors real world Complex relations made simple Queries made simple, very short Shorter application programs No restrictions on data Very efficient full indexing Full indexing -- indexing on every attribute and relationship Flexible classification of objects

Lazy queries Compaction of sparse data No keys are needed Automatic consistency of database Better concurrency control Multi-processor parallelism Interoperability (ODBC, SQL) No tuning required Benchmarks

SEMANTIC DATABASE PRESENTATION OUTLINE

Introduction Definition Features BENEFITS Market Demonstration Summary


10

SEMANTIC DATABASE BENEFITS

Strategic/Enterprise Performance Architecture Development Users

11

SEMANTIC DATABASE BENEFITS-Strategic


FEATURE1 RAPIDLY ADAPTABLE ACCURATELY MODEL THE BUSINESS HIGH AVAILABILITY AND REALIABILITY OFFERS INFORMATION OFFER Complex, Extended, UserDefined and Abstract Data Types
1

STRATEGIC BENEFIT Quickly Meet Changing Business Needs Business Requirements are Inherent in the Design Continual Access to a Priceless Resource

SDB? YES YES

YES

Eliminates Need to Process Data into YES Information Real World Requires these Data Types YES

Source: IDC, 4/98

12

SEMANTIC DATABASE BENEFITS-Strategic


FEATURE1 SUPPORT OO & COMPONENT BASED MODELS SUPPORT COMPLEX DATA RELATIONSHIPS SUPPORT RAPID APPLICATION DEVELOPMENT SUPPORT RAPID ACCOMODATION OF CHANGE
1

STRATEGIC BENEFIT Models of Actual Business Requirements Business Information Intricately Related Fast Time-to-Market

SDB? YES

YES

YES

Market/Business Requirements Change Rapidly

YES

Source: IDC, 4/98

13

SEMANTIC DATABASE BENEFITS-Strategic


FEATURE1 INHERENT SCALABILITY EASY TO USE, ACCESS & ANALYZE
INTEROPERABILITY IN SECURE, DISTRIBUTED, HETEROGENEOUS NETWORKS
1

STRATEGIC BENEFIT

SDB?

New Business Requirements are YES Easily Met The More Information is Used, YES the More Valuable It Becomes Meet Existing Enterprise YES Processing Requirements for Internal, Internet and Electronic Commerce

Source: IDC, 4/98

14

SEMANTIC DATABASE BENEFITS-Strategic


FEATURE2 INSULATE USERS FROM SQL INSULATE USERS FROM RELATIONAL MODEL IMPROVES QUERY PERFORMANCE AND USER SCALABILITY
2

STRATEGIC BENEFIT Eases System Management

SDB? YES

Automates Maintenance of Indices & YES Summaries Reduces Load on Warehouse DBMS YES

Source: Arbor Sotware (Essbase), 4/98

15

SEMANTIC DATABASE BENEFITS-Strategic


FEATURE2 VASTLY ENHANCED BUSINESS CALCULATION CAPABILITY ENABLES READ/WRITE OPERATIONAL OLAP APPLICATIONS SUPPORTS WIDE RANGE OF CLIENT TOOLS
2

STRATEGIC BENEFIT

SDB?

Relieves IT from Generating Reports for Users YES

Enables Central Control Over Analytical Data

YES

Deploys Quickly at Low Risk and Expense

YES

Source: Arbor Software (Essbase), 4/98

16

SEMANTIC DATABASE BENEFITS-Performance SB2 Benchmark Cold Times Times are in Seconds 100,000 Customers

100 90 80 70 60 50 40 30 20 10 0 T1 T2 T3 T4 T5

SDB (82MB) Oracle Compact (141MB) Oracles Sparse (123MB)

17

SEMANTIC DATABASE BENEFITS-Performance SB2 Benchmark Cold Times Times are in Seconds 500,000 Customers

800 700 600 500 400 300 200 100 0 T1 T2 T3 T4 T5 Oracle Compact (648MB) Oracles Sparse (580MB) SDB (304MB)

18

SEMANTIC DATABASE BENEFITS-Performance SB2 Benchmark Hot Times Times are in Seconds 100,000 Customers

T5 T4 T3 T2 T1 0.01 1 100

Oracle Sparse (123MB) Oracle Compact (141MB) SDB (82MB)

19

SEMANTIC DATABASE BENEFITS-Performance SB2 Benchmark Hot Times Times are in Seconds 500,000 Customers

T5 T4 T3 T2 T1 0.01 0.1 1 10 1001000

Oracle Sparse (580MB) Oracle Compact (648MB) SDB (304MB)

20

SEMANTIC DATABASE BENEFITS-Performance TPC-D Benchmark 1GB Times are in Seconds Identical Hardware Published Results for DB2

Q16 Q13 Q10 Q7 Q4 Q1 1 100 10000 DB2 Oracle SDB

21

SEMANTIC DATABASE BENEFITS-Architecture

Semantic Database Facts

Subschema:
COMPANY name: String m:m address: String m:m
manufactures (m:m)

PRODUCT specification: String m:m weight_kg: Number m:m

Facts:

1. object1 COMPANY 2. object1 NAME IBM 3. object1 MANUFACTURES object2 4. object1 MANUFACTURES object3 5. object2 PRODUCT 6. object2 SPECIFICATION Thinkpad 7. object3 PRODUCT 8. object3 SPECIFICATION TrackPoint

Fact types: aC aRy aRv

22

SEMANTIC DATABASE BENEFITS-Architecture

Access Level Basic Queries:


aC aRy a? ?C aR? ?Ra a?+a??+??a ?Rv ?R[v1,v2] Verify the fact aC. Verify the fact aRy. Find all the categories to which a belongs. Find all objects of category C. Find all y such that aRy. Find all abstract objects x such that xRa. Retrieve all the immediate information about an abstract object. Find all x such that xRv. Find all objects x and v such that xRv and v1 < v < v2

1 disk access per basic query

23

SEMANTIC DATABASE BENEFITS-Architecture

Update Transactions Accumulated Transaction: (V,D,I)


New_database=old_database - the-set-of-facts-to-be-Deleted + + the-set-of-facts-to-be-Inserted

V= queries to be verified

24

SEMANTIC DATABASE BENEFITS-Architecture

Direct and Inverted Facts

Direct:
object1 COMPANY object1 NAME IBM object1 MANUFACTURES object2 object1 MANUFACTURES object3 object2 PRODUCT object2 SPECIFICATION Thinkpad object3 PRODUCT object3 SPECIFICATION TrackPoint

Inverted:

COMPANY inv object1 NAME inv IBM object1 object2 MANUFACTURES inv object1 object3 MANUFACTURES inv object1 PRODUCT inv object2 SPECIFICATION inv Thinkpad object2 PRODUCT inv object3 SPECIFICATION inv TrackPoint object3

25

SEMANTIC DATABASE BENEFITS-Architecture

Sorted Facts

Sorted:

COMPANY inv object1 NAME inv IBM object1 PRODUCT inv object2 PRODUCT inv object3 SPECIFICATION inv Thinkpad object2 SPECIFICATION inv TrackPoint object3 object1 COMPANY object1 MANUFACTURES object2 object1 MANUFACTURES object3 object1 NAME IBM object2 PRODUCT object2 MANUFACTURES inv object1 object2 SPECIFICATION Thinkpad object3 PRODUCT object3 SPECIFICATION TrackPoint object3 MANUFACTURES inv object1
26

SEMANTIC DATABASE BENEFITS-Development


Much Shorter Application Development Inherent Data Modeling in the Sem-ODB Minimal Database Design - Sem-ODB is as Users View Their Needs Relations Between Classes/Objects Dont Have to be Programmed Considerably Smaller & Simpler SQL Statements
27

PROJECT name: String key description: String comments: String starting-date: Date ending-date:Date
runs (m:m)

SEMANTIC DATABASE BENEFITS-Development

SCHEMA
serves (m:m)

LOCATION north-UTM: Number key/2 east-UTM: Number key/2 elevation-ft: Number description: String
located at (m:1)

ORGANIZATION is-part-of m:m: name: String key description: String

belongs to (m:m)

PHYSICAL OBSERVATION STATION is-part-of m:1: structure: String comments: String housing: String
by (m:1)

FIXED STATION platform-height-ft: 0..50.000

IMAGE image: Raw subject: String direction-of-view: 0..360 comments: String type: Char(3)

MEASUREMEMENT TYPE name: String key measurement-unit: String upper-limit: Number lower-limit: Number
of (m:1)

OBSERVATION time: Date-time comment: String

MEASUREMENT value: Number


28

SEMANTIC DATABASE BENEFITS-Development


GIVE ME A DESCRIPTION OF ALL ORGANIZATIONS AND THE LOCATION OF THEIR FIXED STATIONS
select description, LOCATION.north-UTM-in-key, LOCATION.east-UTM-in-key from ORGANIZATION, LOCATION where exists ( select * from FIXED-STATION where exists ( select * from PHYSICAL-OBSERVATION-STATIONBELONGS-TO-ORGANIZATION where name-key = organization-name-in-key and PHYSICAL-OBSERVATION-STATIONBELONGS-TO-ORGANIZATION. physical_observation-station-id-in-key = FIXED-STATION.physical-observationstation-id-key and located-at--north-UTM = north-UTM-in-key and located-at--east-UTM = east-UTM-in-key ) )

SQL for RDBMS

SQL for SDB


Select description, LOCATION from ORGANIZATION

29

SEMANTIC DATABASE BENEFITS-Development


GIVE ME ALL OF THE OBSERVATIONS, WITH ALL OF THEIR ATTRIBUTES, SINCE JANUARY 1, 1993, AND THE LOCATION OF THE OBSERVING STATIONS
( select MEASUREMENT-TYPE.*, LOCATION.north-UTM-in-key, LOCATION.east-UTM-in-key, MEASUREMENT.*, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL from MEASUREMENT-TYPE, LOCATION, MEASUREMENT where time > '1993/01' and exists ( select * from FIXED-STATION where by-physical-observation-station-id = physical-observation-station-id-key and located-at--north-UTM = north-UTM-in-key and located-at-east-UTM = east-UTM-in-key and of--name = name-key)) union ( select MEASUREMENT-TYPE.*, NULL, NULL, MEASUREMENT.*, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL from MEASUREMENT-TYPE, MEASUREMENT where time > '1993/01' and not exists ( select * from FIXED-STATION where by-physical-observation-station-id = physical-observation-station-id-key and of-name = name-key)) union ( select NULL, NULL, NULL, NULL, LOCATION.north-UTM-in-key, LOCATION.east-UTM-in-key, NULL, NULL, NULL, NULL, NULL, NULL, IMAGE.* from LOCATION, IMAGE where time > '1993/01' and exists ( select * from FIXED-STATION where by-physical-observation-station-id = physical-observation-station-id-key and located-at-north-UTM = north-UTM-in-key and located-ateast-UTM = east-UTM-in-key)) union ( select NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, IMAGE.* from IMAGE where time > '1993/01' and not exists ( select * from FIXED-STATION where by--physical-observation-station-id = physical-observation-station-id-key))

SQL for RDBMS

SQL for SDB


Select OBSERVATION__, of__, LOCATION from OBSERVATION where time > '1993/01'

30

SEMANTIC DATABASE PRESENTATION OUTLINE

Introduction Definition Features Benefits MARKET Demonstration Summary


31

SEMANTIC DATABASE MARKET


DBMS Market
$12Billion - 2001 (RDBMS) $2 Billion - 2001 (ODBMS)
DBMS Vendor Market Share ($ millions) Informix Oracle 957 IBM Sybase 16 709 630 37 Object Versant

OO-Tools $4Billion DW/DSS $8Billion Internet Market


$21-$32Billion Tools, DBMS & Applications

Sem-ODB
DBMS, Tool, DSS & Internet READY
SOURCES
Corporate Financial Statements IDC Survey

4223

32

SEMANTIC DATABASE MARKET

Sem-ODB Applications:
Internet/WEB Data Access Data Warehouse/DSS Video/Audio/Spatial Data Storage Vertical Applications Geographic Information Systems Visualization/Data Modeling Knowledge Base
33

SEMANTIC DATABASE PRESENTATION OUTLINE

Introduction Definition Features Benefits Market DEMONSTRATION Summary


34

SEMANTIC DATABASE DEMONSTRATION


PROVEN APPLICATIONS: Satellite Observations Databases NASAs EOS 1TB+ per day Everglades National Park Database 300 Classes, 2,500 Attributes 40 Years of Environmental Observations TerraFly Edutainment Control Systems GIS
35

SEMANTIC DATABASE DEMONSTRATION

Visualization: GIS Database: Sem-ODB Size: 1 TB Contents: Landsat Multispectral, USGS Ortho Photography, Ozone Spatial Data, Factual data

36

SEMANTIC DATABASE PRESENTATION OUTLINE

Introduction Definition Features Benefits Market Demonstration SUMMARY


37

SEMANTIC DATABASE SUMMARY

Information Knowledge Base Real World User Accessible High Performance Any Type of Data Cost & Processing Reduction
38

SEMANTIC DATABASE SUMMARY

39

You might also like