Microsoft SQL Server 2008 R2 Unleashed- P212 doc

Say that you want to see sales units and sales returns but across the full product dimension breakouts and full time dimension breakouts for the United States geographic region only.. Th

If you want much more drill-up and drill-down visibility into your data, you could build

up a much more complicated representation in the data browser Say that you want to see

sales units and sales returns but across the full product dimension breakouts and full time

dimension breakouts for the United States geographic region only You also want to see all

dimension levels, totals by levels, and grand totals by dimension You start the same way

as you did earlier and expand out the measures object until you see all the detail measures

in the Comp Sales cube If you still have the previous example in your data browser, you

can simply locate the Clear Results icon in the data browser tab and clear the data browser

pane Then you drag Sales Units to the center of the lower portion of the data browsing

pane (into the Drop Totals or Detail Fields Here section in the lower right) You do the

same for the Sales Returns measure Then you drag the geography dimension to the upper

section called Select Dimensions or just highlight Select Dimensions and choose the

geog-raphy dimension This is the dimension-level filtering capability within the data browser

You now just select (via the drop-downs of each section within a filter specification) the

level and type of filtering you want to do for the dimension you are working with You

can specify any number of filters within any number of dimensions To just filter on

coun-tries within the geography dimension, you select Councoun-tries within the hierarchies list of

the geography dimension, and the operator you want is Equal, and the filter expression is

the data value that you want to filter on (the United States country value, in this case)

These are all drop-down lists that you can easily select by either clicking the entry or

indi-cating which ones to use via a check box entry Figure 51.48 shows the fully specified Geo

Dimension filter specified

FIGURE 51.48 Complex data browsing with full dimensions and filtering in the SMSS data

browser

The data values you now see are only those of the United States You now drag the

product dimension object to the Drop Column Fields Here section (just above where the

data measures were dropped) You immediately see the data measure values being broken

out by the entire product dimension (you expand the plus sign of the product hierarchy

all the way out to the SKU level) Then you drag the time dimension object to the Drop

Row Fields Here section (just to the left of where the data measures were dropped) You

can choose to view the data at any level within either the time or product hierarchies, and

you can filter on any other dimension values You can also just add a dimension or

dimension level to the filter portion within the data browser or just drag off dimensions,

measures, or filters from the data browser if you don’t want to use them anymore This is

very easy indeed The cube browser shows you what your cube has in it and also illustrates

the utility of a dimensional database Users can easily analyze data in meaningful ways

SSAS allows you to browse individual dimension member data You just right-click any

dimension in the left pane of SSMS (for example, the time dimension) and choose Browse

As you can see in Figure 51.49, the dimension browser opens with All as the top node in

the dimension You simply expand the levels to see the actual member values within this

cube dimension Expanding each level gets you to more detailed information as you move

down the dimension hierarchy

FIGURE 51.49 Browsing the Time dimension using SSMS

Delivering Data to Users

SSAS provides a great deal of flexibility for building scalable OLAP solutions, but how do

you present the data to users? The client-side components deliver much of the

functional-ity of SSAS, using the same code base for the dimensional calculation engine, caching,

and query processing You can use the Pivot Table Service to manage client/server

connec-tions, and this is the layer for user interfaces to access SSAS cubes through the OLE DB for

OLAP interface ADO MD provides an application-level programming interface for

devel-opment of OLAP applications Third-party tools and future versions of Microsoft Excel

(like 2007 and 2010) and other Microsoft Office products will use the Pivot Table Service

to access cubes

The underlying Pivot Table Service shares metadata with SSAS, so a request for data on the

client causes data and metadata to be downloaded to the client The Pivot Table Service

determines whether requests need to be sent to the server or can be satisfied at the client

with downloaded data If a user requests sales information for the first quarter of 2008 and

then later decides to query that data for the first quarter of 2007 for comparison, only the

request for 2007 data has to go to the server to get more data The 2008 data is cached on

the client

Slices of data that are retrieved to the client computer can also be saved locally for analysis

when the client computer is disconnected from the network Users can download the data

in which they are interested and analyze it offline The Pivot Table Service can also create

simple OLAP databases by accessing OLE DB–compliant data sources

With the ADO MD interface, developers will be able to access and manipulate objects in

an SSAS database, enabling web-based OLAP application development

Many independent software vendors, such as Brio, Cognos, Business Objects, Micro

Strategies, and Hyperion, are working with Microsoft to leverage the rich features of these

OLAP services They offer robust user interfaces that can access SSAS’s cubes Versions of

Microsoft Office include the Pivot Table Service to enable built-in analysis in tools such as

Excel It is getting easier and easier to bring OLAP to the masses

Multidimensional Expressions

The OLE DB for OLAP specification contains MDX syntax that is used to build datasets

from cubes and is used to define cubes themselves Developers of OLE DB OLAP providers

can map MDX syntax to SQL statements or native query languages of other OLAP servers,

depending on the storage techniques

MDX statements build datasets by using information about cubes from which the data

will be read This includes the number of axes to include, the dimensions on each axis

and the level of nesting, the members or member tuples and sort order of each dimension,

and the dimension members used to filter, or slice, the data (Tuples are combinations of

dimensions such as time and product time that present multidimensional data in a

two-dimensional dataset.)

An MDX statement has four basic parts:

Member scope information, using the WITH MEMBERclause

Dimension, measure, and axis information in the SELECTclause

The source cube in the FROMclause

Dimension slicing in the WHEREclause

Expressions in an MDX statement operate on numbers, strings, members, tuples, and sets

Numbers and strings mean the same thing here as they do in other programming contexts.

Members are the values in a dimension, and levels are groups of members Sets are

collec-tions of tuple elements to further combine facts If the dimension were time, a particular

year, quarter, or month would be a member, and month values would belong to the

month level You use the dimension browser in SSAS to view members of a dimension

The following example shows an MDX SQL expression:

WITH MEMBER [Measures].[Total Sales Units]

AS ‘Sum([Measures].[Sales Units])’

SELECT

{[Measures].[Total Sales Units]} ON COLUMNS,

{Topcount([Product_Dimension].[SKU].members,100,

[Measures].[Total Sales Units])}

ON ROWS

FROM [Comp Sales]

WHERE ([Time_Dimension].[All Time])

You can download this simple query against the Comp Sales cube from Sams Publishing at

www.samspublishing.com, and it is on the CD for this book as well This query returns the

sums of the sales units for products for all time periods Figure 51.50 shows the full

execu-tion of this query within a query window of SSMS Notice that the metadata for the cube

is also made available in the center pane of SSMS, along with an MDX Functions tab that

provides all the MDX functions that can be used This feature is very helpful for building

valid MDS queries within this environment Also notice that the result set display area is

very specialized in order to display multidimensional results

This simple MDX statement shows the basic parts of a working query In this case,

measures are displayed in columns, and the product dimension members make up the

axes of this multidimensional query and are displayed in rows The display of multiple

dimensions in rows like this is how the term tuple is used in the context of SSAS.

Much more could be said about MDX syntax, and a complete discussion of MDX could fill

its own chapter For more information, see the OLE DB for OLAP Programmers Reference,

which is available on the Microsoft website at http://msdn2.microsoft.com/en-us/library/

ms145506.aspx It contains detailed information about MDX expressions and grammar

ADO MD

ADO MD is an easy-to-use access method for dimensional data via an OLE DB for OLAP

provider You can use ADO MD in Visual Basic, Visual C++, and Visual J++ Like ADO,

ADO MD offers a rich application development environment that can be used for

multi-tier client/server and web application development

FIGURE 51.50 Comp Sales MDX query execution in SSMS

You can retrieve information about a cube, or metadata, and execute MDX statements by

using ADO MD to create cellsets to return interesting data to a user ADO MD is another

subject too broad to cover in detail in this chapter Specifications for OLE DB for OLAP

and ADO MD are available on the Microsoft website at http://msdn2.microsoft.com/en-us/

library/ms126037.aspx

Calculated Members (Calculations)

Remember from the Comp Sales requirements that there was an additional user need to

see the difference between sales units and sales returns (sales units minus sales returns) to

yield net sales One approach is to use the SSAS calculated members (calculations)

capabil-ity This creates an expression against existing measures that will be treated the same as a

measure Basically, you need to complete the requirements for the Comp Sales cube by

adding a calculation measure to this cube for net sales units

To create a calculation, you go back to Visual Studio and the cube designer Then you click

the Calculations tab and create a new calculation measure called Sales Units NET with the

calculation expression of (Sales Units - Sales Returns), as shown in Figure 51.51 Many

functions are available for use that should meet your individual calculation needs

This calculation fulfills the data measure requirements of Comp Sales All that is left to do

is to process the cube so others can use it The following sample MDX query uses the

newly created calculation measure:

WITH MEMBER [Measures].[Total Sales Units NET]

AS ‘Sum([Measures].[Sales Units NET])’

FIGURE 51.51 A new calculation measure of Sales Units NET in the Visual Studio cube

designer

Figure 51.52 shows this new calculation measure listed in the cube’s metadata pane You

can see how easy it is to use in the cube data browser You might want to check the math,

however, to make sure the calculation is correct

Query Analysis and Optimization

In SSAS, you can look at query utilization and performance in a cube You can look at

queries by user, frequency, and execution time to determine how to better optimize

aggre-gations If a slow-running query is used frequently by many users, or by the CEO, it might

be a good candidate for individual tuning A usage-based analysis capability can be used

to change aggregations based on actual live queries that the cube must service This

adjusts aggregations based on a query to reduce response time You start this wizard by

right-clicking the cube’s partition Figure 51.53 shows the Usage-Based Optimization

Wizard splash page

The Usage-Based Optimization Wizard allows you to filter queries by user, frequency of

execution, time frame, and execution time You see a record for each query you have run

since the date you began, the number of times it was executed, and the average execution

time, in seconds This is like a SQL trace analysis of your OLAP queries

SELECT

{[Measures].[Total Sales Units NET]} ON COLUMNS,

{Topcount([Product_Dimension].[SKU].members,100,

[Measures].[Total Sales Units NET])}

ON ROWS

FROM [Comp Sales]

WHERE ([Time_Dimension].[All Time])

FIGURE 51.52 Data browsing using the Sales Units NET calculation in the Visual Studio cube

designer data browser

FIGURE 51.53 The Usage-Based Optimization Wizard

Because aggregations already exist, the wizard asks whether you want to replace them or

add new ones If you replace the existing aggregations, the cube is reprocessed with this

particular query in mind

Generating a Relational Database

The examples you have worked with up to this point have been from a dimensional

data-base that uses a star or snowflake schema (theCompSalesdatabase) Very often, however,

you create cubes based on requirements only and do not have an existing data source (or

sources) to draw on at design time After you complete your cube design, you can choose

to generate a relational schema that can be used to retain (that is, stage) the cube’s source

data or that can be a data warehouse/data mart unto itself Figure 51.54 shows the start

of the Schema Generation Wizard for building a data warehouse/staging database from

the top down

FIGURE 51.54 Generating a relational schema from the cube and dimension definitions

NOTE

Designing dimensional databases is an art form and requires not only sound

dimen-sional modeling knowledge, but also knowledge of the business processes with which

you are dealing Data warehousing has several design approaches Regardless of

which approach you take, having a good understanding of the approach’s design

tech-niques is critical to the success of a data warehouse project Although Microsoft

pro-vides a powerful set of tools to implement data marts, astute execution of design

methods is critical to getting the correct data—the truly business-significant business

data—to the end users

Limitations of a Relational Database

Even using a tool such as SSAS, you face limitations when dealing with a normalized

data-base Using a view can often solve (or mask) these issues In some cases, however, more

complicated facts and dimensions might require denormalized tables or a dimensional

database in the storage component of the data warehouse to bring information together

Data cleansing and transformation are also major considerations before you attempt to

present decision makers with data from OLTP systems

Cube Perspectives

A new feature in SSAS is cube perspectives This is essentially a way to create working

views of a complex cube that is focused on just what a particular user or group of users

need They don’t need all the dimensions, calculations, levels, and key performance

indi-cators (KPIs) that would otherwise be visible as part of a complex SSAS cube Therefore,

you need a method to tailor or limit a larger cube environment to be just what the users

need and nothing more—hence, the cube perspective Figure 51.55 shows the Perspectives

tab in the cube designer It allows you to easily customize a view (perspective), which is

what will be deployed or referenced to a target user group In this example, you are

creat-ing a new perspective called Comp Sales wo Sales Price, which excludes the extremely

sensitive Sales Price data measure from any user given access to this perspective

You can have any number of perspectives on a cube Figure 51.56 shows what a cube user

sees when trying to browse (or access) cube data via a perspective

Using perspectives is a great way to simplify the user’s life in an already-complicated

OLAP world

KPIs

Figure 51.57 shows another new capability in SSAS: creating embedded KPIs Just like

calculations, KPIs allow you to define thresholds, goals, status indications, and trend

expressions that become part of an OLAP cube Each can then be graphically displayed in

a variety of ways (for example, gauges, thermometers, traffic lights, trend indications such

as up arrows, smiling faces) This is perfect for an executive dashboard or portal

imple-mentation that has its basis in an SSAS cube You can easily access KPIs via the cube

FIGURE 51.55 Creating cube perspectives within SSAS in the cube designer

FIGURE 51.56 Browsing cube data via a perspective in the cube designer

designer’s KPIs tab What are you waiting for? It is pretty easyto create powerful KPIs with

this simple yet rich interface

Data Mining

With SSAS, a much more robust selection of capabilities for data mining is available

Data mining is the process of understanding potentially undiscovered characteristics or

distributions of data Data mining can be extremely useful for OLAP database design in

that patterns or values might define different hierarchy levels or dimensions that were not