Relational DBMS Support for OLAP(OLAP-3)

8/2/2019 Relational DBMS Support for OLAP(OLAP-3)

1/47

Page 1

Relational DBMS support for

OLAPBy

Dr DVLN Somayajulu

ProfessorDept of CSE

National Institute of Technology

WarangalE-Mail: [email protected]

Relational DBMS support for

OLAPBy

Dr DVLN Somayajulu

ProfessorDept of CSE

National Institute of Technology

WarangalE-Mail: somadvlns@

gmail.com


2/47

Page 2

Lecture 39

y Relational DBMS support for OLAP

y Data Cube Demonstration using SQL


3/47

OLAP Operators

ySlicingyDicing

yDrill downyRoll upyPivot


4/47

OLAP OperatorsOperator Purpose Description

Slice Focus attention on asubset of dimensions

Replace a dimensionwith a single membervalue or with a

summary of itsmeasure values

Allows intuitive manipulation

Imitates Projection operation


5/47

OLAP Operators

Imitates selection operation

Operator Purpose Description




Dice Focus attention on asubset of membervalues

Replace a dimensionwith a subset ofmembers


6/47

OLAP Operators

More general level to detail levelof aggregation







Drill-down Obtain more detail

about a dimension

Navigate from a more

general level to amore specific level


7/47

OLAP Operators

Navigate from more specific level of aggregation

to more general level of aggregation



Replace a dimensionwith a single membervalue or with a summaryof its measure values

Dice Focus attention on asubset of member values


Drill-down Obtain more detail abouta dimension

Navigate from a moregeneral level to a morespecific level

Roll-up Summarize about adimension

Navigate from a morespecific level to a moregeneral level


8/47

OLAP OperatorsOperator Purpose Description


Replace a dimension with asingle member value orwith a summary of itsmeasure values


Replace a dimension with asubset of members

Drill-down Obtain more detailabout a dimension

Navigate from a moregeneral level to a more

specific level

Roll-up Summarize about adimension

Navigate from a morespecific level to a moregeneral level

Pivot Present data in a

different order

Rearrange the dimensions

in a data cube


9/47

Page 9

Simple Aggregation

Query: Add up amounts for day 1In SQL: SELECT sum(amt)

FROM SALEWHERE date =1

Sale ProdId StoreId Date amt

p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8P1 s1 2 44

p1 s2 2 4

81


10/47

Page 10

Simple Aggregation

Query: Add up amounts for day 1In SQL: SELECT sum(amt)

FROM SALEWHERE date =1


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8P1 s1 2 44

p1 s2 2 4

81


11/47

Page 11

Output ofSlice Operator

Location Product

Soda Diet Soda Lime Soda Orange Soda

Hyderabad 80 110 60 25

Kolkata 40 90 50 30

Chennai 70 55 60 35

Delhi 75 85 45 45

Mumbai 65 45 85 60


12/47

Page 12

Dice Operator

y Focus on a subset of membervalues

y Replace dimension with a subset

of valuesy Dice Operation often follows a

slice operation


13/47

Page 13

Result ofDice Operator

40 90 50 30

Soda DietSoda

OrangeSoda

Limesoda

LO

CATION

Kolkata

PRODUCT


14/47

Page 14

Aggregates using SQL Gruoup by Clause

y Add up amounts by dayy In SQL: SELECT date, sum(amt)

FROM SALEGROUP BYdate


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8P1 s1 2 44

p1 s2 2 4


15/47

Page 15

Output of Aggregated Query on Sale table


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8P1 s1 2 44

p1 s2 2 4

ans date sum

1 81

2 48


16/47

Page 16

Other OLAP Operations

y Suitable for hierarchical dimensions Drill-down: add detail to a dimension Roll-up: remove detail from a dimension

Recalculate measure values Pivot: rearrange dimensions


17/47

Page 17

Another Example

y Add up amounts by day, producty In SQL: SELECT date, sum(amt)

FROM SALEGROUP By date, prodId


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8P1 s1 2 44

p1 s2 2 4


18/47

Page 18

Query Result


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8

P1 s1 2 44

p1 s2 2 4


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8

P1 s1 2 44

p1 s2 2 4

Group1

Group2

Group3


19/47

Page 19

Query Result


p1 s1 1 12

p2 s1 1 11

P1 s3 1 50

p2 s2 1 8

P1 s1 2 44

p1 s2 2 4

Sale ProdId Date Amt

P1 1 62

P2 1 19

P1 2 48


20/47

Page 20

Outline

y OLAP Operatorsy Relational DBMS support for OLAPy Data Cube Demonstration in SQLy Categories of OLAP tools


21/47

Page 21

Support for OLAP in DBMS

y Fundamental operation used in DW isaggregation

y Features of SQL were enrichedsupport for performing OLAPoperations in most of commercialDBMSs. Oracle Database provides extensions for

Group by Clause


22/47

Page 22

RelationalDBMSSupport for OLAP

y Group By Grouping Set < column list> Short notation in SQL:99 for series of UNIONed

Queries that are common in reports

y Rollup extension

y Cube extension


23/47

Page 23

Extensions of Group by Clause

y Rollup extension: Compute aggregation such as sum, avg, max,

min, count at increasing levels aggregationincluding highest value (grand total).

y

Cube extension: Computes all possible aggregations Heavy processing load is needed

y Group By Grouping Set Short notation in SQL:99 for series of UNIONed

queries that are common in reports


24/47

Page 24

Outline

y OLAP Operators

y Relational DBMS support for OLAP

y Data Cube Demonstration in SQL

y Categories of OLAP tools


25/47

Page 25

Cross Tab report With Sub Totals

QUERY in SQL:

SELECT channels.channel_desc, countries.country_iso_code,

TO_CHAR(SUM(amount_sold), '9,999,999,999') SALES

FROM sales, customers, times, channels, countries

WHERE sales.time_id=times.time_id AND sales.cust_id=customers.cust_id AND

sales.channel_id= channels.channel_id AND

channels.channel_desc IN ('Direct Sales', 'Internet') AND

times.calendar_month_desc='2000-09 AND

customers.country_id=countries.country_id AND

countries.country_iso_code IN (INDIA',USA')

GROUP BY CUBE(channels.channel_desc, countries.country_iso_code);

Country

India USA Total

Internet 9,597 124,224 133,821

Direct Sales 61,202 638,201 699,403

Total 70,799 762,425 833,224


26/47

Page 26

ROLLUP Extension to GROUP BY

y Extension of Group by clausey Easy to use and Efficienty Syntax: SELECT GROUP BY ROLLUP

(grouping_column_reference_list)


27/47

Page 27

ROLLUP Extension to GROUP BY

y Extension of Group by clausey Easy to use and Efficienty Syntax: SELECT GROUP BY ROLLUP


When to Use ROLLUP?y Use the ROLLUP extension in tasks involving

subtotals.

Suited to perform aggregation of data across hierarchical categories such as timeand horizon.

y For data warehouse administrators using summary tables,ROLLUP can simplify and speed up the maintenance ofsummary tables


28/47

CUBE Extension to Group By

y CUBE takes a specified set of grouping columns

and creates subtotals for all of their possible

combinations.

y CUBE generates all the subtotals that could be

calculated for a data cube with the specifieddimensions.

y CUBE (time, region, department), the result setwill include all the values that would be included inan equivalent ROLLUP statement plus additional

combinations.

y Ifn columns are specified for a CUBE, there will be2 to the n combinations of subtotals returned


29/47

Page 29

Advantages of CUBE Operator

y Data required to generate cross tabreport can easily be computed withsingle SQL statement.

y Very helpful in generating summarytables

y Most suitable for queries suitable formultiple dimensions


30/47

Page 30

CUBE function

Syntax:SELECT GROUP BY CUBE


QUERY:

SELECT channel_desc, calendar_month_desc, countries.country_iso_code,TO_CHAR(SUM(amount_sold), '9,999,999,999') SALES$FROM sales, customers, times, channels, countriesWHERE sales.time_id=times.time_id AND sales.cust_id=customers.cust_id AND

sales.channel_id= channels.channel_id ANDcustomers.country_id = countries.country_id ANDchannels.channel_desc IN ('Direct Sales', 'Internet') AND

times.calendar_month_desc IN ('2000-09', '2000-10') ANDcountries.country_iso_code IN ('GB', 'US')GROUP BY CUBE(channel_desc, calendar_month_desc, countries.country_iso_code ) ;


31/47

Page 31

Partial Cube Extension to

GROUP BYy Partial cube:

Syntax: GROUP BY expr1, cube(expr2, expr3);

- Creates subtotals at 4 aggregate levels

Example Query in SQL

QUERY in SQL:

SELECT channel_desc, calendar_month_desc, countries.country_iso_code,

TO_CHAR(SUM(amount_sold), '9,999,999,999') SALES$

FROM sales, customers, times, channels, countries

WHERE sales.time_id = times.time_id

AND sales.cust_id = customers.cust_id

AND customers.country_id=countries.country_id

AND sales.channel_id = channels.channel_id

AND channels.channel_desc IN ('Direct Sales', 'Internet')

AND times.calendar_month_desc IN ('2000-09', '2000-10')

AND countries.country_iso_code IN ('GB', 'US')

GROUP BY channel_desc, CUBE(calendar_month_desc, countries.country_iso_code);


32/47

Page 32

Discussion on CUBE

y Question: Can we compute SubTotals without using Cube? Multiple Select statements combined with

union operation returns the same result Require many select statements Use of many select statements leads to

inefficient processing and need very lengthSQL statement


33/47

Page 33

Discussion on CUBE

y Question: Can we compute Sub Totalswithout using Cube? Multiple Select statements combined with union operation

returns the same result Require many select statements Use of many select statements leads to inefficient

processing and need very length SQL statement

y Question: What is the impact when we addone more dimension into it?

More attributes/columns are included, we need sixteendifferent aggregation

More dimensions are included, the greater savings withCube operator compared to Unioned approach


34/47

Page 34

GROUPING SETS Expression

y To explicitly specify the set of groups that youwant to create within a GROUP BY clause using aGROUPING SETS expression.

y This allows precise specification across multipledimensions without computing the whole CUBE


35/47

Page 35

Challenges with the use ofRollup and Cube

y How can you programmatically determinewhich result set rows are subtotals,

y How do you find the exact level of

aggregation for a given subtotal?y Is there any way with ease to determine

which rows are the subtotals?

y What happens if query results contain bothstored NULL values and "NULL" valuescreated by a ROLLUP or CUBE?


36/47

Page 36

Lecture 40

Categories of OLAP


37/47

Page 37

Categories of OLAP Tools

y OLAP tools are categorized according tothe architecture of the underlyingdatabase.

y Three main categories of OLAP toolsinclude Multi-dimensional OLAP (MOLAP

or MD-OLAP) Relational OLAP (ROLAP), alsocalled multi-relational OLAP


38/47

Page 38

Multi-Dimensional OLAP (MOLAP)

y Use array technology and efficientstorage techniques that minimize thedisk space requirements through

sparse data management.

y Provides excellent performance whendata is used as designed, and the focus

is on data for a specific decision-support application.


39/47

Page 39

Multi-Dimensional OLAP (MOLAP)

y Traditionally, require a tight couplingwith the application layer andpresentation layer.

y Recent trends segregate the OLAP fromthe data structures through the use ofpublished application programminginterfaces (APIs).


40/47

Page 40

Typical Architecture for MOLAP

Tools


41/47

Page 41

MOLAP Tools - Development Issues

y Underlying data structures are limitedin their ability to support multiplesubject areas and to provide access todetailed data.

y Navigation and analysis of data islimited because the data is designedaccording to previously determinedrequirements.


42/47

Page 42

MOLAP Tools - Development Issues

y MOLAP products require a different setof skills and tools to build and maintainthe database, thus increasing the costand complexity of support.


43/47

Page 43

Relational OLAP (ROLAP)

y Fastest growing style of OLAPtechnology.

y

Supports RDBMS products using ametadata layer - avoids need tocreate a static multi-dimensionaldata structure - facilitates thecreation of multiple multi-dimensional views of the two-dimensional relation.


44/47

Page 44

Relational OLAP (ROLAP)

yTo improve performance, someproducts use SQL engines to

support complexity of multi-dimensional analysis, whileothers recommend, or require,

the use of highly de-normalizeddatabase designs such as thestar schema.


45/47

Page 45

Typical Architecture for ROLAP

Tools


46/47

Page 46

ROLAPROLAP

y A multi-dimensional user view onrelational data storage using Star orSnowflake Database Schemata.

ProductDimension

TimeDimension

RegionDimension

CustomerDimension

ProductDimension

YearDimension

CountryDimension

CustomerDimension

Sales

CustomerCharacteristics

ProductKind

Region

Month

Snowflake Schema

Sales

Star Schema


47/47

Page 47

OLAP TOOLS

y

Organize facts according to multiple dimensionsand they use powerful rules for combining

those facts to form aggregate facts

y Characteristics:

To drill down into the data

To swap the dimensions

Allow changes in the appearance of the data

y Provide flexibility

y Vendors who sell these tools?

Oracle with Oracle Express suite, Cognos withPowerPlay, Microstrategy with DSS agent, and many

others.

Date post:	06-Apr-2018
Category:	Documents
Upload:	sunilpokhrael
View:	233 times
Download:	0 times

Relational DBMS Support for OLAP(OLAP-3)

Documents