Presentasjon lastes. Vennligst vent

Presentasjon lastes. Vennligst vent

WWW.PLATON.NET DENMARK FINLAND ICELAND NORWAY SWEDEN Neste generasjon Datavarehus med SQL Server 2008 Tommy Strandvold 1 Pervasive Insight.

Liknende presentasjoner


Presentasjon om: "WWW.PLATON.NET DENMARK FINLAND ICELAND NORWAY SWEDEN Neste generasjon Datavarehus med SQL Server 2008 Tommy Strandvold 1 Pervasive Insight."— Utskrift av presentasjonen:

1 WWW.PLATON.NET DENMARK FINLAND ICELAND NORWAY SWEDEN Neste generasjon Datavarehus med SQL Server 2008 Tommy Strandvold 1 Pervasive Insight

2 © Platon Microsoft Business Intelligence Vision and strategy 2 Improving organizations by providing business insights to { all } employees, leading to better, faster, more relevant decisions Complete and integrated BI and performance-management offering Widespread delivery of intelligence through Microsoft Office Enterprise-grade and affordable

3 © Platon Deliver Relevant information Drive Actionable Insights Share insights 3 Predictable Response Simplified Management Scale across mixed workloads Integrate your data Connect to any source Develop visually Microsoft SQL Server 2008

4 © Platon4 Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) BI Solutions Data Staging Area (DSA) Data entry Data Profiling Integration Broker The Data Warehouse topology with SQL 2008 Master Data

5 © Platon Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) BI Solutions Data Staging Area (DSA) Extract Data entry Data Profiling Master Data 1 1 Service 1 Integration Broker 1 The Data Warehouse topology with SQL 2008

6 © Platon6 Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) BI Solutions Data Staging Area (DSA) ExtractArchive 2 Data entry Data Profiling 1 1 Service 1 Integration Broker The Data Warehouse topology with SQL 2008 Master Data 1

7 © Platon7 Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) BI Solutions Data Staging Area (DSA) Baseline ExtractArchive 2 Data Errors Data entry Data Profiling 1 1 Service 1 Integration Broker 3 The Data Warehouse topology with SQL 2008 Master Data 1

8 © Platon8 Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) BI Solutions Data Staging Area (DSA) BaselineDMSA 4 ExtractArchive 2 Data Errors Data entry Data Profiling 1 1 Service 1 Integration Broker 3 The Data Warehouse topology with SQL 2008 Master Data 1

9 © Platon9 Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) Data Staging Area (DSA) BaselineDMSA 4 ExtractArchive 2 Data Errors Data entry Data Profiling 1 1 Service 1 Integration Broker 5 5 3 5 The Data Warehouse topology with SQL 2008 BI Solutions Master Data 1

10 © Platon10 BI front-end solution BI front-end solution BI front-end solution BI front-end solution BI front-end solution BI front-end solution BI front-end solution BI front-end solution Source systems Enterprise Data Warehouse (EDW) Data Marts (DM) BI Solutions Data Staging Area (DSA) BaselineDMSA 4 BI front-end solution BI front-end solution ExtractArchive 2 Data Errors Data entry Data Profiling 1 1 Service 1 Integration Broker 5 5 3 Service SQL Service 6 5 The Data Warehouse topology with SQL 2008 Master Data 1

11 © PlatonPage 11 Agenda SQL Server Integration Services SQL Server Database Engine SQL Server Analysis Services

12 © Platon SQL Server Integration Services ●Data Profiling Task og Data Profile Viewer ●ADO.NET source and destination ●Source(nytt navn; het tidligere DataReader Source) ●I SSIS 2005, DataReader Source måtte konfigureres via Advanced Editor og støttet kun SQL skrevet direkte inn i komponenten. ●Destination ●Ny komponent 12

13 © Platon ●Scripting med Visual Studio Tools for Applications ●Fullintegrert IDE med debugging support ●Fremdeles ikke debug i pipeline ●Support for Visual C#.NET ●Mulighet for å referere alle.NET assemblies 13 SQL Server Integration Services

14 © Platon SQL Server Integration Services ●Forbedret Performance og Caching i Lookup Transformation ●3 outputs ●Grønn for lookup match, grønn for lookup failure og rød for komponent error ●Cache Mode ●Bedre performance med precompiled scripts ●Alle script må nå være pre-compiled! ●Pipeline Performance ●Paralell prossesering ●Annet ●Output config i Message window ●Ny @[System::ParentContainer] variabel 14

15 © PlatonPage 15 Agenda SQL Server Integration Services SQL Server Database Engine SQL Server Analysis Services

16 © Platon SQL Server Relational DBMS ●MERGE(Upsert) INSERT INTO edw.DimProsjekt(ParentDimKey, DimAlternateKey, fraDato) SELECT prosjektId, prosjektKode, getdate() FROM ( MERGE edw.DimProsjekt as EDW_prosjekt USING extract.Prosjekt AS src_delta ON (EDW_prosjekt.dimkey = src_delta.ProsjektId and EDW_prosjekt.TilDato is null) WHEN MATCHED THEN UPDATE SET EDW_prosjekt.TilDato = getdate() WHEN NOT MATCHED THEN INSERT VALUES (src_delta.prosjektId, src_delta.Prosjektkode, getdate()) OUTPUT $action, src_delta.prosjektId, src_delta.prosjektKode ) AS Changes(action, prosjektId, prosjektKode) WHERE action = 'UPDATE'; ●Intellisence ●Endelig!!!! ●Change Data Capture ●Hente endringer i et gitt tidsrom/siden sist ●Er det en UPDATE, INSERT eller DELETE som har skjedd ●Hvilke kolonner har UPDATE kjørt på 16

17 © Platon ●Resource Governor CREATE FUNCTION rgclassifier() RETURNS SYSNAME WITH SCHEMABINDING AS DECLARE @grp_name AS SYSNAME IF (SUSER_NAME() = 'sa’) SET @grp_name = 'groupAdmin' IF (APP_NAME() LIKE '%Management Studio%’) SET @grp_name = 'groupAdhoc' IF (APP_NAME() LIKE '%Analysis Services%’) SET @grp_name = 'groupASProsessing' RETURN @grp_name ●Star Join Query (Enterprise Edition) ●Dynamisk bruk av Bitmap Filter/Bloom Filter Algoritme www.wikipedia.org/wiki/Bloom_filter www.wikipedia.org/wiki/Bloom_filter 17 SQL Server Relational DBMS

18 © Platon ●Data Compression ●ROW compression ●fixed datatyper blir konvertert til variable length ●PAGE compression ●column prefix og page level dictionary ●Backup Compression ●Minimally logged INSERT 18 SQL Server Relational DBMS

19 © PlatonPage 19 Agenda SQL Server Integration Services SQL Server Database Engine SQL Server Analysis Services

20 © Platon SQL Server Analysis Services ●Attribute Relationship Designer ●Enhanced Dimension and Aggregation Designers ●Best Practice Design Alerts ●BIDSHelper ●www.codeplex.com/bidshelperwww.codeplex.com/bidshelper 20

21 © Platon BIDS Demo 21 SQL Server Analysis Services

22 © Platon ●CREATE KPI ●Definerer KIP’er i MDX script eller som en Session KPI ● CREATE KPI CURRENTCUBE | Cube_Name.KPI_Name AS KPI_Value [, FORMAT_STRING=string, VISIBLE=value, NON_EMPTY_BEHAVIOR=, CAPTION=string, DISPLAY_FOLDER=string, ASSOCIATED_MEASURE_GROUP= ] ●UPDATE MEMBER ●Endre deler av en kalkulert medlem uten å endre alle egenskaper ● UPDATE MEMBER Cube_Name.Member_Name as mdx_expression [, Property_Name = Property_Value,…n] 22 MDX Syntax Extensions

23 © Platon MDX Syntax Extensions ●CREATE MEMBER ● CREATE [ SESSION ] [HIDDDEN] [ CALCULATED ] MEMBER CURRENTCUBE | Cube_Name.Member_Name AS MDX_Expression [,Property_Name = Property_Value,...n] [, CAPTION=string, DISPLAY_FOLDER=string, ASSOCIATED_MEASURE_GROUP= ]......[,SCOPE_ISOLATION = CUBE] ●CREATE SET ● CREATE [SESSION] [STATIC | DYNAMIC][HIDDEN] SET CURRENTCUBE | Cube_Name.Set_Name AS 'Set_Expression' [, CAPTION=string, DISPLAY_FOLDER=string] 23

24 © Platon Static Versus Dynamic Sets ●Static Set ●Kalkuleres ved evaluering av MDX’en ●Tar ikke hendsyn til subcube/where ●Dynamic Set ●Kalkuleres i kontekst av subcube/where 24

25 © Platon ●Scale out Analysis 25 SQL Server Analysis Services

26 © Platon ●Block Computation ●Kalkulerer og beregner kun non-null verdier i en MDX spørring ●Celle navigering gjøres kun en gang ●PrevMember, LastMember, FirstChild, Parent osv. ●Block Computation er verdt oppgradering til SQL 2008 alene ●Har størst effekt på kuber med mye null verdier 26 SQL Server Analysis Services

27 © Platon Cell by Cell Computation 27 Measures.[Sales Amount] Measures.[Product Contribution] (Measures.[Sales Amount], Product.[Product Categories].Parent / =

28 © Platon Block Computation Logic ●Fakta ●A = Product.CurrentMember ●B = Product.Parent ●Bare når A er not null, A / B er not null ●B er konstant ●Storage Engine henter bare non-null data ●Konklusjon ●Kjører kun kalkulering når teller-verdiene returneres fra Storage Engine og nevner har en konstant verdi 28

29 © Platon Block Computation 29 Measures.[Sales Amount] Measures.[Product Contribution] (Measures.[Sales Amount], Product.[Product Categories].Parent Steg 1: Hent non-null verdier fra Storage Engine Steg 2: Kalkuler Steg 3: Legg til null verdier i resultatet

30 © Platon Oppsummering ●Bedre ytelse i ETL ●SQL 2008 gjør det mye enklere å ha et real-time datavarehus enn tidligere ●Bedre AS ytelse med Block Computation ●Enklere å forsikre seg om at kuben er optimal 30

31 © Platon31


Laste ned ppt "WWW.PLATON.NET DENMARK FINLAND ICELAND NORWAY SWEDEN Neste generasjon Datavarehus med SQL Server 2008 Tommy Strandvold 1 Pervasive Insight."

Liknende presentasjoner


Annonser fra Google