DW Implementation Boot Camp - Student Manual

5/22/2018 DW Implementation Boot Camp - Student Manual

1/118

For Oracle employees and authorized partners only. Do not distribute to third parties. 2008 Oracle Corporation Proprietary and Confidential


The following is intended to outline our generalproduct direction. It is intended for informationpurposes only, and may not be incorporated into anycontract. It is not a commitment to deliver anymaterial, code, or functionality, and should not berelied upon in making purchasing decisions.The development, release, and timing of anyfeatures or functionality described for Oraclesproducts remains at the sole discretion of Oracle.

Safe Harbor Statement


Use of this Site (Site) or Materials constitutes agreement with the following terms and conditions:

1. Oracle Corporation (Oracle) is pleased to allow its business partner (Partner) to download andcopy the information, documents, and the online training courses (collectively, Materials") found on thisSite. The use of the Materials is restricted to the non-commercial, internal training of the Partnersemployees only. The Materials may not be used for training, promotion, or sales to customers or otherpartners or third parties.

2. All the Materials are trademarks of Oracle and are proprietary information of Oracle. Partner or otherthird party at no time has any right to resell, redistribute or create derivative works from the Materials.

3. Oracle disclaims any warranties or representations as to the accuracy or completeness of anyMaterials. Materials are provided "as is" without warranty of any kind, either express or implied,including without limitation warranties of merchantability, fitness for a particular purpose, and non-infringement.

4. Under no circumstances shall Oracle or the Oracle Authorized Delivery Partner be liable for any loss,damage, liability or expense incurred or suffered which is claimed to have resulted from use of this Siteof Materials. As a condition of use of the Materials, Partner agrees to indemnify Oracle from and againstany and all actions, claims, losses, damages, liabilities and expenses (including reasonable attorneys'fees) arising out of Partners use of the Materials.

5. Reference materials including but not limited to those identified in the Boot Camp manifest can not beredistributed in any format without Oracle written consent.

Oracle Training Materials Usage

Agreement


Title of Presentation

Presenters NamePresenters Title

Data Warehouse course introduction


2/118


Module Agenda

Introductions

Introduction to data warehousing

Creating business value from IT

Data warehouse architecture & design

Preparing for the hands-on labs


Introductions

The goals of this course

Provide background to understand technicalchallenges presented by data warehouses

Provide a technical overview of the Oracle Database11g Data Warehousing features with extensivehands-on labs

Guidelines on sizing of Data Warehouse hardwareconfigurations

Share best practices for deploying Oracle DataWarehouse architectures and platforms


Introductions

Who are you? Name

Company

Role

What is your prior experience? Database administration

Data warehouse experience

10g or 11g Database experience

How do you expect to benefit from this course?



3/118

Introductions

Who am I?


Course modules

Introduction to data warehousing

Introduction to Oracle data warehouse features

Parallel execution - Lab

Partitioning - Lab

Result Cache - Lab

OLAP - Lab

Advanced compression - Lab

Data integration and ETL- Demo

Course modulesContinued

Sizing Oracle data warehouse platforms

Exadata overview

Best practices for performance


Introduction to data

warehouses


4/118


All about data warehousesWhat is a data warehouse?

A data warehouse is A repository of data

From multiple sources

Used for analysis

Strategic value is realized from data aggregation


All about data warehousesWhy have a data warehouse?

Increase the value inherent in data

Extract strategic information from tactical data

Drive IT value into business results


All about data warehousesHow is a data warehouse different from OLTP?

Read-mainly

Typically, real-time not required

Large data sets

Complex queries based on many factors


All about data warehousesHow is a data warehouse the same as OLTP?

Modern data warehouses Near real time

Mixed workloads

ETL

OLTP

Write operations make OLTP issues relevant

Performance

Locking


5/118


All about data warehousesUsing a data warehouse

Data

WarehouseCustomers Web

Servers

Personalization

Content

Servers

Fulfillment

Centers

Supply

Chain

Planning

Apps

Financial

Analysis

Campaign

Generation Email

Servers

Internal

End Users

Partners


All about data warehousesUsing a data warehouse Amazon.com

0

100

200

300

400

500

600

700

Terabytes

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

NA

FE

EU

Data Volume has grown 2x year over year for the past 4 years.Projecting higher growth in upcoming years*:

- additional lines of business / product lines supported- huge standard reporting growth with more partners supported


Creating business value

from IT


IT PRIORITIES BUSINESS PRIORITIES

Reduced Cost Business Value

Improved Security Ease of Access

Improved SystemPerformance

Results in Timefor Deadlines

Manageability /Consistency

Good Enough Infrastructure



6/118


Increasing Return on Investment for informationproducers by Reducing current costs of analysis for lines of business

Empower LOBs with tools, applications, and automatedanalysis rather than manually creating custom reports

Reducing ongoing costs of analysis for IT

Faster and simpler development of analytic applications

Centralized management



Increasing Return on Investment for informationconsumers in lines of business by Enabling top-line growth

New customers/markets, best customers,

React faster than the competition

Enabling bottom-line savings

Optimize operations, mitigate risk,

Better manage the business than the competition



Data warehouse

architecture and design


Data warehouse architecture and designArchitecture

Data warehouse

Operational data store

Data marts

Federated data marts


7/118


Data warehouse architecture and designData design

Analysis performed on a set of facts Examined through different dimensions

Usually has time dimension

Queries typically examine facts with many dimensions


Data warehouse architecture and designDesign options

Star schema

Snowflake schema

OLAP cubes

Hybrid organization


Data warehouse architecture and designDesign options Star schema

SalesSales

DistributorDistributorOEMOEM

ChannelChannel Geography Geography

TypeType

SizeSize

ColorColor

ProductProduct TimeTime

YearYear

QuarterQuarter

MonthMonth

RegionRegion

DistrictDistrictBranchBranch

Fact Table:Fact Table:

SalesSales

TransactionsTransactions


Data warehouse architecture and designDesign options Snowflake schema


8/118


Data warehouse architecture and designDesign options OLAP cubes

Product

Sales Cube

Inventory Cube

SALES FACTSALES FACT

TIME

INVENTORYINVENTORY

REGION

CUST

ITEM

TimeGeography

Product

Time


Data warehouse architecture and designDesign options Hybrid organization

Combination of structures Star schema

OLAP cubes

Third normal form

Oracle query optimizer can handle appropriately forall structures


Lab Setup

Lab setup - Connection

Instructions for dialing in to hosted image



9/118

Lab setup - Testing

Login to the VM Linux environment with oracle/oracle

Open a new command window

Start a SQL Plus session as SH/SH


Lab setup VM environment overview

All passwords for OS & Oracle logins are same as theusername

Except - root, SYS, SYSTEM, and SYSMAN areoracle

All the lab scripts are stored in the/home/oracle/wkdirdirectory

Few shortcuts:

Command

Window

Firefox

Browser

Analytic

Workspace

Manager

SQL

Developer


Summary

A data warehouse is a way to provideadditional value from existing data

Data warehouse architectures can includeoperational data stores, federated datamarts and data marts

Data warehouses can use star schemas,snowflake schema, OLAP cubes, thirdnormal form or a mixture of differentstructures




10/118

For Oracle employees and authorized partners only. Do not distribute to third parties.

2008 Oracle Corporation Proprietary and ConfidentialFor Oracle employees and authorized partners only. Do not distribute to third parties.

2008 Oracle Corporation Proprietary and Confidential




Data Warehouse and Business Intelligenceoverview


Module Agenda

Why Oracle for data warehousing?

Core data warehousing features

Oracle 11g data warehousing features

Oracle hardware data warehouse solutions

Analytics and data mining

Oracle Business Intelligence Platform (OBIEE)


11/118


Why Oracle for data

warehousing?

Why Oracle for data warehousing?

Best-of-breed for data warehouses and data marts

Consolidation platform Eliminates the costs and inefficiencies of multiple data stores

Complete, integrated ELT and data-qualitycapabilities

Unmatched analytic capabilities Embedded OLAP, data-mining and statistics all accessible via

SQL

Optimized for grid computing

Next-generation BI with integrated OLTP and DSS

Optimized hardware solutions

Foundation of Oracle Business Intelligence solutions


Oracle - #1 in data warehouse market

share


Core data warehousing

features


12/118

Oracle data warehouse platform

DW Services

Administration

Workload Mgmt

Self-Tuning

Security

Virtual PrivateDatabase

Database Vault

Data Integration

Bulk ETL + Real-TimeETL

Data Quality

Metadata ManagementExtensibleRepository

Lineage and ImpactAnalysis

Brainy Software

Analytics

Multi-DimensionalCalculations

Statistics Data Mining

Intelligent Query Processing

MaterializedViews

BitmapIndexes

PartitionElimination

Star QueryOptimization

Brawny Hardware

Scalable Data Management

ParallelExecution

Partitioning RACAutomatic

StorageMgmt

Compression

Scalable Hardware Infrastructure

ReferenceConfigurations

OptimizedWarehouse

ExadataDatabaseMachine

ExadataStorage Server


Core data warehouse featuresBit-mapped indexes

PARTStable 001001 GREENGREEN MEDMED 98.198.1

002002 REDRED MEDMED 12411241003003 REDRED SMALLSMALL 100.1100.1004004 BLUEBLUE LARGELARGE 54.954.9005005 REDRED MEDMED 124.1124.1006006 GREENGREEN SMALLSMALL 60.160.1...... ........ .......... ......

sizesize == SMALLSMALL 0 0 1 0 1 1 0 1 0 1 0 0 0 1 0 10 0 1 0 1 1 0 1 0 1 0 0 0 1 0 1sizesize == MEDMED 1 0 0 0 0 0 1 0 0 0 0 1 0 1 0 01 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0sizesize == LARGELARGE 0 1 0 1 0 0 0 0 1 0 1 0 1 0 1 00 1 0 1 0 0 0 0 1 0 1 0 1 0 1 0

Index on COLOR

colorcolor == BLUEBLUE 0 1 0 1 0 0 1 0 1 0 1 0 0 0 1 00 1 0 1 0 0 1 0 1 0 1 0 0 0 1 0colorcolor == REDRED 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 10 0 1 0 0 0 0 1 0 0 0 0 1 0 0 1colorcolor == GREENGREEN 1 0 0 0 1 1 0 0 0 1 0 1 0 1 0 01 0 0 0 1 1 0 0 0 1 0 1 0 1 0 0

010

010

010

010

partnopartno color color sizesize weightweight

Index on SIZE

SELECT count(*)

FROM parts

WHERE

size = MED a AND

color = RED


Core data warehouse featuresParallel bit-mapped star joins

Dramatic performance gains

Innovative use of bitmap indexes

Ideal for complex star schemas Multiple fact tables

Many dimensions

Unconstrained dimensions

Large dimension tables

Optimized for sparse fact tables

Parallel execution

Optimizer modifies an SQL query for fasterperformance with star schema

Bitmapped indexes used to select rows from facttables Replaces multiple joins

Results are joined to dimension tables for additionalinformation

Must enable in database and conform to constraints

Core data warehouse featuresStar transformation


13/118


Core data warehouse featuresMaterialized views

What were salesWhat were salesin the West andin the West and

South regions forSouth regions for

the last threethe last three

quarters?quarters?

x

(2 GB)

SALESSALESSALESSALES

SALESSALESPERPER

MONTHMONTHBY REGIONBY REGION

SALESSALESPERPER

MONTHMONTHBY REGIONBY REGION

Stored summaries aremaintained for fasterwarehouse queryprocessing


Core data warehouse featuresMaterialized views query rewrite

SELECT p.brand, r.country, t.month,SUM(s.amt) tot_sales

FROM sales s, region r,time t, product p

WHERE s.region_id = r.region_idAND s.sdate = t.curdateAND s.prod_code = p.prod_codeGROUP BY p.brand, r.country, t.monthHAVING SUM(s.amt) > 5000000;

SELECT brand, country, month, tot_salesFROM sales_sumryWHERE tot_sales > 5000000;

Optimizer

transparently

rewrites queryon detailed

data ...

Optimizer

transparently

rewrites queryon detailed

data ...

to access

data in the

materialized

view

to access

data in the

materialized

view


Core data warehouse featuresMaterialized views data refresh

Maintain the timeliness of storedresult to preference

Several options for refreshing datain materialized views

Manage the effect of performanceon operational data

No impact on load performance

SUMSUMORDERSORDERSbyby

QUARTERQUARTER

COUNTORDERS

byPRODUCT

AVGORDERS

byREGION

InsertInsert

UpdateUpdate

DeleteDelete

InsertInsert

UpdateUpdate

DeleteDelete

DeferredDeferredRefreshRefresh

FullFullRefreshRefresh

ORDERSORDERS

IncrementalIncrementalRefreshRefresh


Core data warehouse featuresParallel execution

select c.cust_last_name

, sum(s.amount_sold)

from customers c, sales s

where c.cust_id = s.cust_id

group by c.cust_last_name;

Data on Disk Parallel Servers

scanscan

scanscan

scanscan

aggregateaggregate

Scanners

Coordinator

joinjoin

joinjoin

joinjoin

aggregateaggregate

aggregateaggregate

Joiners Aggregators

RAC aware!


14/118


Core data warehouse featuresPartitioning

Core functionality Performance Manageability

Oracle8 Range partitioning

Global range indexes

Static partitionpruning

Basic maintenanceoperations: add,drop, exchange

Oracle8i Hash and compositerange-hash partitioning

Partition-wise joins

Dynamic pruning

Merge operation

Oracle9i List partitioning Global indexmaintenance

Oracle9iR2 Composite range-listpartitioning

Fast partition split

Oracle10g Global hash indexes Local Indexmaintenance

Oracle10gR2 1M partitions per table Multi-dimensionalpruning

Fast drop table

OracleDatabase 11g

More composite choicesREF PartitioningVirtual Column Partitioning

Interval PartitioningPartition Advisor


Core data warehouse featuresData compression

Tables and indexes can be compressed Can be specified on a per-partition basis

Typical compression ratio 4:1

Requires more CPU to load data Decompression minimal resource cost

Compress for all DML operations

Less data on disk Requires less time to read

Little or no impact on query performance

Completely transparent

Especially applicable to data warehouses

4XCompression


Core data warehouse featuresData compression real world examples

0

500

1000

1500

2000

2500

Table Scans

0

0.1

0.2

0.3

0.4

2.5x Faster

DML Performance

0

10

20

30

40

< 3% Overhead

3x Saving

Data Storage


Core data warehouse featuresDatabase Resource Manager

High PrioritySales Analysis20 users

(DOP 10)

Medium PriorityAd Hoc Reports200 users

(DOP 4)

Low PriorityETL Jobs

200 users

(DOP 4)

Protect the system pro-actively Maximum number of concurrent operations

Control CPU and IO consumption based on Consumer Groups


15/118


Core data warehouse featuresSQL Access Advisor

Based on workload

Materialized view advice

Index advice

Partition advice


Core data warehouse featuresSQL Access Advisor how it works

Indexes Material izedviews

Materializedviews log

SQL AccessAdvisor

Hypothetical

SQL cache

FilterOptions

STS

CompleteWorkload

Partitionedobjects

Hashpartitions?

Intervalpartitions?


Core data warehouse featuresOracle Warehouse Builder

Design Target WarehouseDesign Target Warehouse1

2 Define SourcesDefine Sources

Relational Flat Files Applications Mainframe

Map Source to TargetMap Source to Target3

Oracle or flat files

Generate CodeGenerate Code4

Instantiate WarehouseInstantiate Warehouse5

6 Extract & transform data


Core data warehouse featuresOracle Warehouse Builder - Repository

Multi-user repository Single update, multi-read

Object locking

Synchronize

Metadata import / export &interchange Bridges to Oracle tools, ERwin,

PowerDesigner

Bridges to other metadata availablefrom Metadata Integration Technology

Incorporated


16/118


Core data warehouse featuresOracle Data Integrator

Next Generation Architecture

E-LTE-LTLoadExtract

Transform Transform

Conventional ETL Architecture

Extract LoadTransform Performance

Heterogeneous E-LT

Productivity Declarative design

Flexibility Active integration design

Easily extendable Knowledge modules


Core data warehouse featuresReal Application Clusters

QueryQueryQueryQueryETL ETL ETL ETL

Data Warehouse

Scale across cluster



QueryQueryETL ETL

Data Warehouse


Dedicate nodes tospecific work



QueryQueryETL ETL

Data Warehouse


Dedicate nodes tospecific work

Scale out whenworkload increases

Query


17/118


Spreads database files acrossall available storage for optimalperformance and resourceutilization

Enables simple, online resourceallocation and providesautomatic data rebalancing

Up to 3-way mirroring

Can leverage external hardwareRAID mirror and parity protection

Core data warehouse featuresAutomatic Storage Management (ASM)

Oracle OLAP OptionEmbedded, manageable, enterprise-ready

A multidimensional calculation and aggregation engine Multidimensional data types: Cubes and dimensions

An OLAP API for cube definition and multidimensional queries

A SQL interface to OLAP Cubes and dimensions

Embedded in the Oracle Database Runs within Oracle instance on same server

OLAP cubes are stored in Oracle data f iles

OLAP metadata in the Oracle Data D ictionary

All key database functionality extends to OLAP Security: Object and data security

Scalability: Real Application Clusters

Availability: Backup-Recovery, Disaster Recovery, RAC

Administration: Enterprise Manager


Why Oracle OLAP is so FastData Access Method

Data stored in dense arrays

Offset addressing no joins

More powerful analysis

Better performance

Time

Product

CD

ProfitDVD

Audio

Q1 Q2 Q3SF

West

USA

Market

Compare Profit this Quarter versus Last Quarter

What is a Products Profit Share of its Category?



Core data warehouse featuresCurrent adoption

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

DB Res Mgr

RMAN

ASM

Read Only

VPD

MV Use

Compression

Parallel Exec

Partitioning


18/118


Oracle 11g data

warehousing features


Oracle 11g data warehouse featuresData Result Cache

Automatically caches results of queries, query blocks, or PL/SQLfunction calls

Cache is shared across statements and sessions on serv er

Significant speed up for read-only / read-mostly data

Full consistency and proper semantics

Cache refreshed when any underlying table updated

join

join

Table1 Group-by

join

Table2 Table3

Q2: Uses resulttransparentlycache


Oracle 11g data warehouse featuresData Results Cache

Retail customer data (~50 GB) Concurrent users submitting queries randomly

Executive dashboard with 12 heavy analytical queries

Cache results only at in-line view level

12 queries run in random, different order 4 queries cached

Measure average, total response time for all users

447 s

267 s

186 s

No cache

334 s

201 s

141 s

Cache

25%

25%

24%

Improvement

8

4

2

# Users


Oracle 11g data warehouse featuresSQL PIVOT/UNPIVOT

Rotate rows into columns and vice versa Common operation for crosstab queries

PROD QUARTER AMOUNT

Shoes Q1 2000

Shoes Q2 1000

Jeans Q1 1000

Jeans Q3 500

Jeans Q3 100Jeans Q4 1000

PROD Q1 Q2 Q3 Q4

Shoes 2000 1000 Null Null

Jeans 1000 Null 600 1000

SELECT * FROM tPIVOT (sum(amount)

FOR quarter inQ1,Q2,Q3,Q4);


19/118


Oracle 11g data warehouse featuresAdditional features

Enhanced partitioning options

Advanced compression

OLAP-based materialized views

Priority-dependent maximum degree of parallelism inDatabase Resource Manager


Oracle data warehousing

hardware

CustomCustom

Complete Flexibility

Any OS, any platform

Easy fit into acompanys ITstandards

Oracle DW Hardware SolutionsThe evolution of hardware infrastructure


CustomCustom



Easy fit into acompanys ITstandards

Documented best-practiceconfigurations fordata warehousing






20/118

CustomCustom



Easy fit into acompanys IT

standards


Optimized

Warehouse

Optimized

Warehouse

Scalable systemspre-installed and pre-

configured: ready torun out-of-the-box

Reference

Configurations

Reference

Configurations



CustomCustom



Easy fit into acompanys IT

standards


Optimized

Warehouse

Optimized

Warehouse

Scalable systemspre-installed and pre-

configured: ready torun out-of-the-box

Highest performance

Pre-installed and pre-configured

Sold by Oracle

Reference

Configurations

Reference

ConfigurationsSun Oracle

Database

Machine

Sun Oracle

Database

Machine



Build From Scratchwith Components

OWI ReferenceConfigurations

Take delivery of OWI

configuration

Weeks to Months

Pre-implementationSystem sizing

Acquisition ofcomponents

Installation andconfiguration

Acquisition ofcomponents

Installation andconfiguration

Testing andValidation

Testing andValidation

Weeks to Months

Sun OracleDatabase Machine

Faster deploymentLower Risk

< 1 Week

Database Machine BenefitsAccelerate Implementations & Lower Risk



Analytics and data

mining


21/118

SQL analytics

Bring Algorithms to the Data - Not Data tothe Algorithms

Analytic computationsdone in the database SQL analytics

Statistics

OLAP

Data Mining

Scalability

Security

Backup & Recovery

Simplicity

OLAP

Data Mining

Statistics


Why Data Mining?Do you wish to ..

Identify factors most associated with a business problem?

Predict customer behavior ?

Predict or estimate a value ?

Find profiles of targeted people or items?

Segment a population ?

Determine important relationships/market baskets within thepopulation ?

Find fraudulent or rare events ?

(Attribute Importance)

(Classification)

(Regression)

(Decision Trees)

(Clustering)

(Associations)

(Anomaly Detection)


Statistics & SQL Analytics

Ranking functions rank, dense_rank, cume_dist, percent_rank, ntile

Window Aggregate functions(moving and cumulative)

Avg, sum, min, max, count, variance, stddev,first_value, last_value

LAG/LEAD functions Direct inter-row reference using offsets

Reporting Aggregate functions Sum, avg, min, max, variance, stddev, count,

ratio_to_report

Statistical Aggregates Correlation, linear regression family, covariance

Linear regression Fitting of an ordinary-least-squares regression line

to a set of number pairs.

Frequently combined with the COVAR_POP,COVAR_SAMP, and CORR functions.

Descriptive Statistics average, standard deviation, variance, min, max, median (via

percentile_count), mode, group-by & roll-up

DBMS_STAT_FUNCS: summarizes numerical columns of atable and returns count, min, max, range, mean, stats_mode,variance, standard deviation, median, quantile values, +/- nsigma values, top/bottom 5 values

Correlations Pearsons correlation coefficients, Spearman's and Kendall's

(both nonparametric).

Cross Tabs Enhanced with % statistics: chi squared, phi coefficient,

Cramer's V, contingency coefficient, Cohen's kappa

Hypothesis Testing Student t-test , F-test, Binomial test, Wilcoxon Signed Ranks

test, Chi-square, Mann W hitney test, Kolmogorov-Smirnovtest, One-way ANOVA

Distribution Fitting Kolmogorov-Smirnov Test, Anderson-Darling Test, Chi-

Squared Test, Normal, Uniform, Weibull, Exponential

Pareto Analysis 80:20 rule, cumulative results table


Oracle Data Mining OptionEmbedded, manageable, enterprise-ready

A complete data-mining solution 10+ data-mining algorithms: addresses every business scenario

Generalized Linear Models (Logistic & Multiple Regression)

Oracle Data Miner: Graphical user-interface for analysts

A Data Mining PL/SQL & Java API for developers

Embedded in the Oracle Database Runs within Oracle instance on same server

Data Mining algorithms execute directly on database t ables

Data Mining models and metadata stored within Oracle database

All key database functionality extends to Data Mining Scalability: Parallelism, Real Application Clusters

Security: Object and data security

Administration: Enterprise Manager Portability: All major platforms



22/118

Oracle Data MiningOracle in-Database Mining Engine

Oracle Data Miner (GUI) Simplified, guided data mining

Spreadsheet Add-In forPredictive Analytics 1-click data mining from a

spreadsheet

PL/SQL API & Java (JDM) API Develop advanced analytical

applications

Wide range of algorithms



Oracle Business

Intelligence

OLTP & ODSSystems

DataWarehouseData Mart

Oracle Business Intelligence EEData warehouse components


OLTP & ODSSystems


SAP, OraclePeopleSoft, Siebel,

Custom Apps

FilesExcelXML

BusinessProcess

Oracle Business Intelligence EEFull data foundation



23/118

OLTP & ODSSystems



Custom Apps

FilesExcelXML

BusinessProcess

Oracle BI Server

Common Enterprise Information Model

Oracle Business Intelligence EECommon information model


OLTP & ODSSystems



Custom Apps

FilesExcelXML

BusinessProcess

Oracle BI Server


Reporting&

Publishing

Ad hocAnalysis

ProactiveDetectionand Alerts

MicrosoftOffice

InteractiveDashboards

Advantages:

Consolidate and

standardize BI

tools

Unified BI

metadata and

infrastructure

Seamless BI userexperience

Intelligent

Caching Services

Data Federation

Intelligent Alerting

Oracle Business Intelligence EEComprehensive, integrated BI suite


Oracle BI Enterprise Edition Plus

(OBIEE+)

OLTP & ODSSystems



Custom Apps

FilesExcelXML

BusinessProcess

PerformanceManagementApplications

Insight

PerformanceAction

SetGoals

Plan

Monitor

Analyze

Report

Align

HyperionEssbase

Oracle BI Server


Reporting & Publishing

Financial Reporting

BI Publisher

Interactive Reporting

SQR Production Reporting

Web Analysis

Ad hocAnalysis

ProactiveDetectionand Alerts

MicrosoftOffice

InteractiveDashboards

Extend

ODBC


Summary

Oracle has many features to enhance datawarehouse operations

Oracles data warehouse features areimplemented transparently to developers

Oracles data warehouse features areintegrated with the Oracle database

Oracle has evolved the hardwareinfrastructure of data warehouses topreconfigured optimized warehouses

Oracle offers a complete set of analytic andbusiness intelligence capabilities



24/118








Parallel execution


25/118


Module Agenda

Parallel execution overview

Parallel execution setup

Parallel joins

Parallel execution with RAC


Parallel execution

overview

What is parallel execution?

With parallel execution, a task is Broken into smaller sub-tasks, which are executed independently in

parallel, with the result joined at completion


SELECT c.cust_last_name, sum(s.amount_sold)FROM customers c, sales s

WHERE c.cust_id = s.cust_idGROUP BY c.cust_last_name;

Data on Disk Parallel Servers

scanscan

scanscan

scanscan

aggregateaggregate

Coordinator

joinjoin

joinjoin

joinjoin

aggregateaggregate

aggregateaggregate


SQL Parallel Execution

QC is theQC is the user session that initiates the parallel SQL statementuser session that initiates the parallel SQL statement& it will& it will distribute the work to parallel serversdistribute the work to parallel servers

Parallel serversParallel servers -- individual sessions that performindividual sessions that perform

work in parallel They are allocated from awork in parallel They are allocated from a pool ofpool ofglobally available parallel server processes andglobally available parallel server processes andassigned to a given operationassigned to a given operation


26/118

Dynamic parallelismAdvantages

Uses granules

Advantages Uses optimal queueing technique

Avoids convoy effect

Adaptable to changing conditions

Structural

Runtime

Allow partitioning based on business requirements


Parallel executionParallel scans

Data is partitioned into granules(block range or partition) Dynamic Granules Computed

during runtime by the database

Each scan parallel process isassigned multiple granules

No two scanners ever contendfor the same granule

Granules are assigned so thatthe load is balanced across allparallel scan processes

PQ #2

PQ #3

PQ #1

Parallel ExecutionServers

Parallel executionIntra- and inter-operation parallelism

PQ 1

PQ 2

PQ 3

PQ 4

Sales tableFull ScanParallel Sort

A - C

D - F

G - M

N - O

PQ 5

PQ 6

PQ 7

PQ 8

S

T - V

W - Z

P - R

SELECT ... FROM product p, sales s where p.prod_id = s.prod_id ORDER BY p.prodname;

UserProcess

QueryCoordinator

Intra-Parallelism Intra-Parallelism

DOP = 8(Intra-Parallelism)

DOP = 16(Inter-Parallelism)

Inter-Parallelism


Parallel executionParallel scans of multiple tables

SELECT c.cust_name, s.date,

s.amount

FROM sales s, customers c

WHERE s.cust_id= c.cust_id;

Querycoordinator

SALES

Table

CUSTOMERS

Table


27/118


28/118


Parallel executionParallel scans of multiple tables

SELECT c.cust_name, s.date,

s.amount

FROM sales s, customers c

WHERE s.cust_id= c.cust_id;

Querycoordinator

P1 P2 P3 P4

SALES

Table

CUSTOMERS

Table

Assembly

Degree of Parallelism = number of parallel execution serversassociated with a parallel operation

Assigned for database, object, session or query DOP is calculated at run-time

Default DOP (when not specified) =CPU_COUNT xPARALLEL_THREADS_PER_CPU

DOP may be reduced depending on the availability of parallelexecution servers and system load

Database Resource Manager can throttle DOP per resourceconsumer groups

Parallel executionDegree of Parallelism (DOP)

DOP applies to intra-operation parallelism In inter-operation pallelism, parallel execution servers

can be twice the number specified by the DOP No more than two sets of parallel execution servers

can be used for one parallelized statement at a time

Parallel executionDegree of Parallelism (DOP) - continued

SQL Access Methods Table Scans

Fast Full Index Scans

Partitioned Index Range Scans Sorts, Aggregations, Set Operations

External Table access Joins

Nested Loop, Sort-Merge Hash, Star, Partition-Wise

DDL CTAS, CREATE INDEX, REBUILD INDEX

DML INSERT-SELECT, UPDATE, DELETE, MERGE

Parallel executionParallelizable operations


29/118


Parallel execution setup

Parallel Execution Server Pool Server pool size is determined by PARALLEL_MIN_SERVERS

and PARALLEL_MAX_SERVERS Dynamic based on demand

Adaptive Multiuser and DOP PARALLEL_ADAPTIVE_MULTI_USER = TRUE (default) Throttles DOP to prevent overloading of server

PARALLEL_MIN_PERCENT is the minimum parallel executionservers requested for the operation

Default 0 Disabled Operation not executed if the MIN % of DOP servers not available

Parallel execution setupConfiguration

Enabling Parallel DML, DDL and QueryALTER SESSION (ENABLE | DISABLE | FORCE) PARALLEL

(DML | DDL | QUERY) (PARALLEL n);

FORCEPARALLEL overrides default DOP Hints overrides a forced DOP

Parallel Query & Parallel DDL is enabled if ANY ofbelow is TRUE

Object was created with PARALLEL clause ALTER SESSION [ENABLE|FORCE]PARALLEL Parallel hints are specified in the query

Parallel DML enabled if both are TRUE ALTER SESSION [ENABLE|FORCE]PARALLEL Object was created with PARALLEL clause OR Parallel

Hints are specified in DML

Parallel execution setupExecution


Paralleljoins


30/118

Serial Join

scan scan

join

Products Sales

Sales.prod_id = products.prod_id

Scan of both tablesin serial

Compares (joins) allrecords from saleswith products

Basic Parallel Join

scan scan

join

scan scan

join


Scan of both tablesin parallel


Products Sales

Parallel Join with Broadcasting

scan scan

join

scan

join


Scan of both tablesin parallel/serial


Products Sales

Smaller table isbroadcast to all

processes

Parallel Join with Broadcasting

scan scan

join

scan

join


Scan of both tablesin parallel/serial


Smaller table isbroadcast to allprocesses

No data distributionfor large table

Products Sales


31/118

Parallel processesData Redistribution Methods

HASH Standard redistribution for parallel processing

Example JOIN, GROUP BY operations

KEY Ensures the logical clumping of KEY values

RANGE Used for combined GROUP BY and ORDER BY operations

BROADCAST Sends smaller data sets to all participating processes to

optimize data shipping

ROUND ROBIN Final operation to send data to requesting process

Variance LOCAL as RAC optimization

Partition-wise Join

scan scan

join


Products Sales



Equi-partitioning oftables will beleveraged

Partition-wise Parallel Join

scan scan

join

scan scan

join


Products Sales




No data distributionrequired


Parallel execution with

RAC


32/118

Parallel Query

PQ 1

PQ 2

PQ 3

PQ 4


A - C

D - F

G - M

N - O

PQ 5

PQ 6

PQ 7

PQ 8

S

T - V

W - Z

P - R


UserProcess

QueryCoordinator

Internode Parallel Query

PQ 1

PQ 2

PQ 3

PQ 4


A - C

D - F

G - M

N - O

PQ 5

PQ 6

PQ 7

PQ 8

S

T - V

W - Z

P - R


UserProcess

QueryCoordinator

Node 1

Node 2

Many large-scale DWs have many concurrent jobs

Multiple small-to-medium size queries

Degree of parallelism < CPUs-per-node

With Oracle, queries will automatically run on a single node,eliminating traffic over the interconnect

RAC and Parallel Execution

Q1 Q2 Q4Q3

Q5 Q7Q6 Q8

Q9 Q12Q11Q10

Very large queries utilize all resources on thecluster

RAC and Parallel Execution

Large Query


33/118

RAC and parallel executionAllocating parallel execution servers

If no DOP has been set for the object, session orthrough a hint, RAC uses parallel servers from acrossall nodes.

Else, if DOP < 2 times the number of CPUs on anode, RAC will first attempt to use parallel executionprocesses on owner node

If DOP is too high or if enough local parallel serverprocess are not available, processes spread overmultiple nodes

P1 P1 P2 P2 P3 P3 P4 P4

Node 1 Node 2

PX1 PX2 PX3 PX4

Internode Parallel QuerySample optimizations

Classical co-located partition-wise join A partition-wise parallel join is used if the DOP is either equal to

the number of partitions, or an even divisor of the number ofpartitions

If there were 16 partitions, a partition-wise parallel join wouldwork with a DOP of 16, 8, 4 or 2

Example shown 2 nodes, 4 partitions on 2 tables, DOP 4

P1 P1 P2 P2 P3 P3 P4 P4

Node 1 Node 2

Internode Parallel QuerySample optimizations

Without partition-wise parallel join, data redistributionmust occur, which can stress interconnect Sample join of two tables, each with 4 partitions, on 2 nodes,

DOP 8

PX1 PX2 PX3 PX4 PX5 PX6 PX7 PX8

Parallel Operations with RACUsing services

PARALLEL_INSTANCE_GROUPS andINSTANCE_GROUPS nolonger used to controlparallelism across RACnodes

Services specify whichnodes can participate inparallel execution

Connection of a processto a service throughsql*net can contain theparallel execution

Define service thru OEM

or (DB and srvctl) thenadd to tnsnames.ora


34/118

Summary

Oracle uses dynamic parallelism

The degree of parallelism is dynamicallycalculated, based on several factors

Parallel partition-wise joins are theperformance goal

Parallelism is properly supported with RAC

Instance groups or services can limitparallelism to specific nodes with RAC







35/118




Partitioning


Module Agenda

Partitioning benefits

Partitioning options

Index partitioning

Partition operations


Partitioning benefits

Dividing a table or index into smaller portions,based on a partition key

Partitioning transparent to applications and users

PartitioningDefinition



36/118

Partition pruning

Optimizer eliminates unneeded partitions for data access

Improves query performance by several orders (FTS .vs. FPS)

Prunes range or list partitioning for LIKE, equality, and IN-list

Prunes hash partitioning for equality and IN-list

Static and dynamic partition pruning

Partition-wise join

Used when two tables are joined together and when both the tables arepartitioned on the join key

Breaks the large join into smaller joins that occur between each of thepartitions

Significantly reduces response time and improves the use of both CPU andmemory resources

Partitioning benefitsPerformance


Maintenance operations can address smallerpartitions rather than full table or index

Maintenance operations like DROP, SPLIT,COALESCE make it easy to adjust partitions

Partitions support a 'rolling window load or purgeprocess

Partitioning benefitsManageability


Partitioned database objects provide partitionindependence

Important part of a high-availability strategy

Reduce scheduled downtime

Partitioning benefitsAvailability


Partitions can be stored on different storage tiers

Archived partitions can be stored on cheaper storagedevices

Partitioning benefitsCost



37/118


Partition options

Range

Hash

List

Partitioning optionsClassic


Partitioning optionsComposite partitioning

Table SALESRANGE(order_date)-RANGE(ship_date)

Jan 2006

... ...

Feb 2006 Mar 2006 Jan 2007

... ...

... ...

...

...

...

...

Jan2006

Feb2006

Jan2007

Data ispartitioned alongtwo dimensions(A,B)

A distinct valuepair for the twodimensionsuniquelydetermines thetarget partitioning



Table SALESRANGE(order_date)-RANGE(ship_date)

Jan 2006

... ...

Feb 2006 Mar 2006 Jan 2007

... ...

... ...

...

...

...

...

Jan2006

Feb2006

May2006

Mar 2006

All records withorder_date inMarch 2006ANDship_date inMay 2006

May2006



38/118

Composite partitioning options

11g11g11gList

8i9iR211gRange

HashListRange



Partitioning optionsInterval partitioning

Extension to range partitioning

Full automation for equi-sized range partitions

Partitions are created automatically as data arrives No need to create new partitions

Local indexes are created and maintained as well


Table SALES

Jan 2006

... ...

Feb 2006 Mar 2006 Jan 2007 Oct 2009 Nov 2009

...

Interval partitioned table can have classical range andautomated interval section Automated new partition management plus full partition

maintenance capabilities:Best of both worlds



Table SALES

Jan 2006

... ...

Feb 2006 Mar 2006 Jan 2007 Oct 2009 Nov 2009

...


maintenance capabilities: Best of both worlds

MERGE and move old partitions for ILM

Range partition section




39/118

Table SALES

Jan 2006 Feb 2006 Mar 2006 Jan 2007 Oct 2009 Nov 2009


maintenance capabilities:Best of both worlds

MERGE and move old partitions for ILM

Range partition section

... ... ...

Interval partition section

VALUES ('13-NOV-2009')Insert new data- Automatic segment creation



Table SALES

...

2005 Q1 2006 Oct 2006

Range partitioned tables can be extended into intervalpartitioned tables Simple metadata command

Investment protection

Q2 2006

...



Table SALES

...

2005 Q1 2006 Oct 2006Q2 2006

ALTER TABLE sales (order_date DATE, ...)SET INTERVAL(NUMTOYMINTERVAL(1,'month');

Range partitioned tables can be extended into intervalpartitioned tables Simple metadata command

Investment protection

...

New monthlyInterval partitions

Old range partition table

...



Related tables benefit from same partitioning strategy Sample 3NF order entry data model

Redundant storage of the same information solvesthis problem Data overhead

Maintenance overhead

Business Problem

Solution Oracle Database 11gintroduces REF Partitioning

Child table inherits the partitioning strategy of parent tablethrough PK-FK relationship

Intuitive modelling Enhanced performance and manageability

Partitioning optionsREF partitioning



40/118

Table ORDERS

Jan 2006

... ...

Feb 2006

Table LINEITEMS

Jan 2006

... ...

Feb 2006

RANGE(order_date)

Primary key order_id

RANGE(order_date)

Foreign key order_id

Redundant storage of order_date

Redundant maintenance

Partitioning optionsBefore REF partitioning


Table ORDERS

Jan 2006

... ...

Feb 2006

Table LINEITEMS

Jan 2006

... ...

Feb 2006

RANGE(order_date)

Primary key order_id

RANGE(order_date)


RANGE(order_date)


PARTITION BY REFERENCE Partitioning key inherited through

PK-FK relationship

Partitioning optionsWith REF partitioning


Want to create extended schema attributes are fullyderived and dependent on existing common data

Redundant storage or extended view definitions aresolving this problem today Requires additional maintenance and creates overhead

Business Problem

Solution Oracle Database 11g introduces virtual columns

Purely virtual, meta-data only

Treated as real columns except no DML Virtual columns can have statistics

Virtual columns are eligible as partitioning key Enhanced performance and manageability

Partitioning optionsVirtual column-based partitioning


ORDERS

ORDER_ID ORDER_DATE CUSTOMER_ID...---------- ----------- ----------- --9834-US-14 12-JAN-2007 659208300-EU-97 14-FEB-2007 396543886-EU-02 16-JAN-2007 45292566-US-94 19-JAN-2007 153273699-US-63 02-FEB-2007 18733

REGION requires no storage

Partition by ORDER_DATE, REGION

REGION AS (SUBSTR(ORDER_ID,6,2))------

USEUEUUSUS

Partitioning optionsVirtual column-based partitioning



41/118


Index partitioning

Created as equi-partition with theunderlying table; created with theLOCAL attribute

the indexpartition key ranges andtable partition key ranges areidentical

Managed automatically by server toensure equi-partitioning

Easier administration

Enables parallelism of index scans

Reduction of scheduled downtime

Only one index partition affectedby maintenance operation on anunderlying table partitionPartition

Q1

Partition

Q2

Partition

Q3Partition

Q4

SALES07 Table

IndexPartition

IndexPartition

IndexPartition

IndexPartition

CREATE INDEX order_idxON sales07 (order_no)

LOCAL(PARTITION INDEX_Q1,PARTITION INDEX_Q2 ,PARTITION INDEX_Q3,PARTITION INDEX_Q4);

CREATE INDEX order_idxON sales07 (order_no)

LOCAL(PARTITION INDEX_Q1,PARTITION INDEX_Q2 ,PARTITION INDEX_Q3,PARTITION INDEX_Q4);

Local partitioned indexes


41

Not equi-partitioned with theunderlying table

The index is partitioned by the indexkey, not by the table partition key

OLTP applications may getbetter performance becausethey minimize the number ofindex probes

Harder to manage than localindexes

Can be unique or non-uniquePartitionQ1

Partition

Q2

Partition

Q3Partition

Q4

SALES TablePartitioned by Quarter

CREATE INDEX product_idxON sales07 (product_id)

PARTITION BY RANGE(product_id)GLOBAL(PARTITION PIDX1 VALUES LESS THAN (1000),PARTITION PIDX2 VALUES LESS THAN (2000),PARTITION PIDX3 VALUES LESS THAN (MAXVALUE));

CREATE INDEX product_idxON sales07 (product_id)

PARTITION BY RANGE(product_id)GLOBAL(PARTITION PIDX1 VALUES LESS THAN (1000),PARTITION PIDX2 VALUES LESS THAN (2000),PARTITION PIDX3 VALUES LESS THAN (MAXVALUE));

Index PartitionProducts

0-999


1000-1999


2000 +

Global partitioned indexes


All index data within one schemaobject

Treated the same as global indexes

Administration advantages gainedwhen table is partitioned are lost ifassociated indexes are notpartitioned

Why? If recovering a partition of a table,it will only be usable once the entire non-partitioned Index is recovered and usable

Non-Partitioned Index

Partition

Q1

Partition

Q2

Partition

Q3Partition

Q4

SALES07 Table

CREATE INDEX sales07_idxON sales97 (sales_date);

CREATE INDEX sales07_idxON sales97 (sales_date);

Non-partitioned indexes



42/118


Partition operations

Partition-wise join

scan scan

join


Products Sales


Compares (joins) allrows from sales withproducts



Partition-wise parallel join

scan scan

join

scan scan

join


Products Sales




No data distributionrequired with RAC


Rolling window operations

Q106 Q206 Q306 Q406

Order Table(partitioned by quarter)

Drop

Other data & queries not affected

Q107

Add



43/118

EXCHANGE PARTITION

The basic technique of bulk-loading new data into atemporary load table, which is then indexed, analyzed,and then published all at once to end-users using theEXCHANGE PARTITION operation

Should be the default load technique for all large tables ina data warehouse


22-Feb

2008

23-Feb

2008

24-Feb

2008

(empty)

Composite-partitioned

table TXN

2. Bulk

Loads

5. EXCHANGE PARTITION

3. Create

Indexes

4. Table

/ Col Stats

EXCHANGE PARTITION

1. Create

Temp Table


25-Feb

2008

Hash-partitioned

table TXN_TEMP

Summary

Partitioning improves performance andmanageability

Oracle supports many types of partitioning,including range, list, hash, interval, virtualcolumn and reference partitions

Oracle supports composite partitions

Performance goal is partition-wise joins

Indexes can be local or global

Partition exchange can be used to add

data to a warehouse with very highavailability




44/118







Result Caches


Module Agenda

SQL Result Cache

PL/SQL Function Cache


45/118


SQL Result Cache

Analyze data across large data sets Reporting

Forecasting & trend analysis

Data mining

Use parallel execution for good performance

Result Very I/O intensive workload direct reads from disk

Memory is less important

PGA memory more sensitive based on sort or aggregations

SQL Result CacheData warehouse workloads


SQL Result CacheData warehouse sample query


Accesses many rows

Returns few rows

select p.prod_category, sum(s.amount_sold) revenuefrom products p, sales swhere s.prod_id = p.prod_idand s.time_idbetween to_date('01-JAN-2006','dd-MON-yyyy')and to_date('31-DEC-2006','dd-MON-yyyy')

group by rollup (p.prod_category)

SQL Result CacheWhat it does


Caches results of queries, query blocks, orPL/SQL function calls

Read consistency is enforced DML/DDL against dependent database objects invalidates

cache

Bind variables parameterize cached result withvariable values


46/118

SQL Result CacheHow it works


Table 1

Table 2 Table 3

join

join

Group by

query 1executes



Table 1

Table 2 Table 3

join

join

Group by

Table 1

Table 2 Table 3

join

join

Group bycachedresult

result iscached



Table 1

Table 2 Table 3

join

join

Group bycachedresult

Table 5 Table 6

join

Table 4

join

Group by

joinQuery 2 uses cachedresult transparently

RESULT_CACHE_MODE initialization parameter MANUAL, use hints to populate and use

FORCE, queries will use cache without hint

RESULT_CACHE_MAX_SIZE initialization parameter Default is dependent on other memory settings

(0.25% of MEMORY_TARGET or 0.5% of SGA_TARGETor 1% of SHARED_POOL_SIZE)

0 disables result cache

Never > 75% of shared pool (built-in restriction)

/*+ RESULT_CACHE */ hint in queries

SQL Result CacheEnabling SQL Result Cache


47/118

SQL Result CacheSQL hint example

Use RESULT_CACHE hint

select /*+ RESULT_CACHE */ p.prod_category, sum(s.amount_sold) revenuefrom products p, sales swhere s.prod_id = p.prod_idand s.time_idbetween to_date('01-JAN-2006','dd-MON-yyyy')and to_date('31-DEC-2006','dd-MON-yyyy')

group by rollup (p.prod_category)

SQL Result CacheView hint example

Use RESULT_CACHE hint in view definitionselect prod_subcategory, revenuefrom(select /*+ RESULT_CACHE */ p.prod_category, p.prod_subcategory, sum(s.amount_sold) revenuefrom products p, sales swhere s.prod_id = p.prod_idand s.time_idbetween to_date('01-JAN-2006','dd-MON-yyyy')and to_date('31-DEC-2006','dd-MON-yyyy')

group by rollup (p.prod_category, p.prod_subcategory))where prod_category = 'Women'/

Similar query for different prod_category can use cached result

SQL Result CacheExecution plan for query using result cache

------------------------------------------------------------------| Id | Operation | Name |------------------------------------------------------------------| 0 | SELECT STATEMENT | || 1 | RESULT CACHE | fz6cm4jbpcwh48wcyk60m7qypu || 2 | SORT GROUP BY ROLLUP | ||* 3 | HASH JOIN | || 4 | PARTITION RANGE ITERATOR| ||* 5 | TABLE ACCESS FULL | SALES || 6 | VIEW | index$_join$_001 ||* 7 | HASH JOIN | || 8 | INDEX FAST FULL SCAN | PRODUCTS_PK || 9 | INDEX FAST FULL SCAN | PRODUCTS_PROD_CAT_IX |------------------------------------------------------------------

Depends... based on Query repetitiveness

Query execution times

DML activity (cache invalidation frequency)

Remember data warehouse workload Query may run 30 minutes

Query may return 5 rows

Query served from result cache would take a split second

SQL Result CacheOpportunity for benefit


48/118

Use DBMS_RESULT_CACHE package to manage

Use V$RESULT_CACHE_* views to monitor

SQL Result CacheManagement

Result cache is disabled for queries containing Temporary or dictionary tables

Non-deterministic PL/SQL functions

Sequence CURRVAL and NEXTVAL

SQL functions current_date, sysdate, sys_guid, etc.

Result cache for distributed queries Set RESULT_CACHE_REMOTE_EXPIRATION > 0

0 means distributed queries are not cached

Default is 0

DML/DDL on remote database will not expire cached results

SQL Result CacheRestrictions

Result cache does not automatically release memory Grows until maximum size is reached

DBMS_RESULT_CACHE.FLUSH purges memory

Bind variables Cached result is parameterized with variable values

Cached results can only be found for same variable values

Cached result will not be built if Query is built on a non-current version of data (read consistency

enforcement)

Current session has outstanding transaction on table(s) in query

Flashback queries can be cached

SQL Result CacheCaveats

Retail customer data (~50 GB)

Concurrent users submitting queries randomly Executive dashboard application with 12 heavy analytical queries

Cache results only at in-line view level 12 queries run in random, different order

4 queries benefiting from the cache

Measure average, total response time for all users

# Users No cache Cache Im provem ent

2 186 s 141 s 24%

4 267 s 201 s 25%

8 447 s 334 s 25%

SQL Result CacheInternal benchmark


49/118


PL/SQL Function Cache

Calculate a complex derived metric like the ratio ofthe highest median income grouped by state to thelowest median income grouped by state over thewhole population

You need a PL/SQL function

Like the results stored in the SQL Result Cache, thedata changes fairly slowly but the query is repeatedfairly often

PL/SQL Function CacheThe challenge

function f2 return t1%rowtype

is...

beginselect a, m into r1.a, r1.b from ...;

select a, m into r2.a, r2.b from ...;

r.a := r1.a + r2.a;r.b := r1.b + r2.b;return r;

end f2;

~ 2,000 milliseconds for each new call

PL/SQL Function CacheFunction implementation

function f2 return t1%rowtyperesult_cache relies_on(t1, t2)

is...

beginselect a, m into r1.a, r1.b from t1;

select a, m into r2.a, r2.b from t2;

r.a := r1.a + r2.a;r.b := r1.b + r2.b;return r;

end f2;

~ 0 milliseconds for each new call

PL/SQL Function CacheUsing the function cache


50/118

Both are cross-session and RAC interoperable Set RESULT_CACHE_MAX_SIZE > 0 on all instances, or = 0

on all instances

Memory pool is instance-specific

Both build on the same infrastructure Same RESULT_CACHE_MAX_SIZE initialization parameters

Same DBMS_RESULT_CACHE management package

Same V$RESULT_CACHE_*performance views

PL/SQL Function CacheSimilarities with SQL Result Cache

Summary

The Result Cache saves results of queries,reducing overhead needed on repeateduse

Result cache is invalidated if there is DMLor DDL on cached tables

Result Cache can be used to satisfy part ofa query

PL/SQL Function Cache can cachefunction results






51/118





Oracle OLAP


Module Agenda

Oracle OLAP Option

Cube Organized Materialized Views

Querying OLAP Cubes


Oracle OLAP Option


52/118

A full featured multidimensional OLAPserver

Excellent query performance for ad-hoc /unpredictable query

Enhances the analytic content of businessintelligence application

Fast

Documents

DW Implementation Boot Camp - Student Manual