Relational OLAP and Aggregate Navigators

24
Relational OLAP and Aggregate Navigators Data Warehousing Lab. M.S. 2 Hyeyoung Cho

description

Relational OLAP and Aggregate Navigators. Data Warehousing Lab. M.S. 2 Hyeyoung Cho. OLAP. OLAP(OnLine Analytical Processing) One technology for querying the warehouse Suited for summarizing and analyzing huge quantities of data 4 basic functions Multidimensionality Drill down - PowerPoint PPT Presentation

Transcript of Relational OLAP and Aggregate Navigators

Page 1: Relational OLAP and Aggregate Navigators

Relational OLAP and Aggregate Navigators

Data Warehousing Lab.M.S. 2 Hyeyoung Cho

Page 2: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP

OLAP(OnLine Analytical Processing) One technology for querying the warehouse Suited for summarizing and analyzing huge q

uantities of data 4 basic functions

Multidimensionality Drill down Rotation Multiple modes of view

Page 3: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

Multidimensionality provide data about performance

measures, broken down by one or more dimensions

dimensions combine to describe measures vary according to industry, strategy, kinds of

information that systems capture

filtered by dimension and/or measure value

Page 4: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

Page 5: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

Drill down break a summary item into its detailed

components multiple parallel hierarchies

DATE dimension(year vs. seasonal year)

both share a common atomic data element : day

Page 6: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

view highly summarized data and then navigate down to less summarized data

drill across:drill down into a different dimension

rolling up:drill from a detail back up to a total

Page 7: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

Rotation

rotation

month on the Y-axisproduct on the X-axis

month on the X-axisproduct on the Y-axis

Page 8: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

change view perspective drill down and rotation

Page 9: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP basic functions

Multiple modes of view View data in a variety of

formats(graph,chart,etc.)

Page 10: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP database technologies

OLAP application for querying and viewing data regardless of how that data is stored

ROLAP and MOLAP ROLAP (Relational OLAP) :RDB기반 MOLAP (Multidimensional OLAP) :MDDB기반

data marts tuned for specific subject areas more normalized than simple star schema

Page 11: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

OLAP database technologies

Cube(dataset to analyze) three dimension :

height,width,depth(dice) Comprised of any number of

dimensions

Page 12: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

MDDB

Array RDB : store data in tables and columns MDDB : store data in large multidimensional arrays

Multidimensional Database Vendors Oracle Express Server(OES) Essbase(Hyperion Software) Powerplay(Cognos) Gentia(Gentia Software)

Suitable for analyzing narrowly focused sets of data

Page 13: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

RDB

Aggregate Navigation Oracle 8i ’s query rewrite capabilities Materialized views in Oracle 8i

ease the construction and maintenance of the table

speed up query performance

Page 14: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

aggregate navigators select the best table for each query know which summaries exist and the size

RDB

Page 15: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs. Online Report Writing support

Generate SQL calls, Format the results not support

OLAP basic function, Aggregate navigation ROLAP tool vendors

Oracle Discoverer MicroStrategy DSS Agent, Computer Associates DecisionBase

RDB

Page 16: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs.MOLAP

query performance query response time

precalculate all combinations and summaries of data

three dimensions: 1 개의 3D atomic level array 3 개의 2D atomic level array 3 개의 1D summary array broken by each dimension

ROLAP MOLAP

hard to predict performance

predictable, fast response to queryprecalculate possible values in the hypercubes!

Page 17: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs.MOLAP

Page 18: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs.MOLAP

load performance populate data structures and perform

calculations timeROLAP MOLAP

loaded more quickly, refresh refreshed on a daily basis not build all possible summaries!

long load time, not preferable for frequent updates, refreshed on a monthly basisprecalculating data takes time!

Page 19: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs.MOLAP

Analytic capability

ROLAP MOLAP

Sometimes hampered by the limitations of SQL(ignore the limitations,put intermediate results into temporary database tables,use three-tier architectures)Oracle 8i provides terrific SQL extensions(cube,rollup,etc.)!

better support for time series and statistical analysis

Page 20: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs.MOLAP

Dataset Sizes

Dimension Handling

ROLAP MOLAP

not incur sparsity penaltiesvirtually unlimited growth

Grow in size very rapidlySparsity:null, Add dimensions(doubles in size)precalculated summary valuesPhysical limitations

ROLAP MOLAP

construct star schemas and query, summarize, drill down on any of dimension columns

not provide such flexibility with dimensionsproblem of database size explosions as dimensions are added!

Page 21: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP Vs.MOLAP

Maintenance Effort

ROLAP MOLAP

more effort to populate and maintainmultiple structures must be filledindexes and constraints need to be turned on or off during the processAdditional indexes and summary tables may need to be created

strong in the area of maintenance: self-maintaining

Page 22: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP and MOLAP Harmony

Which technology wins? relational database : a large, cross-functional, e

nterprise data warehouse multidimensional database : a well-defined, high

ly targeted analysis-focused data mart limited dimensionality and little need for detailed, atomic-level data

A corporate data warehouse feeds smaller, narrowly focused or stand alone data marts→ ROLAP and MOLAP are complementary!

Page 23: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

ROLAP and MOLAP Harmony

drill through integrate ROLAP and MOLAP

technologies drill down in the MDDB until reach in the

lowest level of detail query to RDB that contains very detailed

atomic-level data not perfect yet so require the custom code

each drill-through query

Page 24: Relational OLAP and Aggregate Navigators

LaboratoryData Warehousing

Conclusion

OLAP is a user interface, not a data storage, concept

OLAP basic functions: multidimensionality, Drill down, Rotation, Multiple modes of view

ROLAP Vs. MOLAP What is a best approach to OLAP?