Kimball Dimensional Modeling Techniques – Part II

Ralph Kimball introduced the data warehouse/business intelligence industry to dimensional modeling in 1996 with his seminal book, The Data Warehouse Toolkit. Since then, the Kimball Group has extended the portfolio of best practices. Drawn from The Data Warehouse Toolkit, Third Edition (coauthored by Ralph Kimball and Margy Ross, 2013), here are the “official” Kimball dimensional modeling techniques, part II.

filo01_Cartoonized_4Basic Dimension Table Techniques

Dimension Table Structure

Every dimension table has a single primary key column. This primary key is embedded as a foreign key in any associated fact table where the dimension row’s descriptive context is exactly correct for that fact table row. Dimension tables are usually wide, flat denormalized tables with many low-cardinality text attributes. While operational codes and indicators can be treated as attributes, the most powerful dimension attributes are populated with verbose descriptions. Dimension table attributes are the primary target of constraints and grouping specifications from queries and BI applications. The descriptive labels on reports are typically dimension attribute domain values.

Dimension Surrogate Keys

A dimension table is designed with one column serving as a unique primary key. This primary key cannot be the operational system’s natural key because there will be multiple dimension rows for that natural key when changes are tracked over time. In addition, natural keys for a dimension may be created by more than one source system, and these natural keys may be incompatible or poorly administered. The DW/BI system needs to claim control of the primary keys of all dimensions; rather than using explicit natural keys or natural keys with appended dates, you should create anonymous integer primary keys for every dimension. These dimension surrogate keys are simple integers, assigned in sequence, starting with the value 1, every time a new key is needed. The date dimension is exempt from the surrogate key rule; this highly predictable and stable dimension can use a more meaningful primary key.

Natural, Durable, and Supernatural Keys

Natural keys created by operational source systems are subject to business rules outside the control of the DW/BI system. For instance, an employee number (natural key) may be changed if the employee resigns and then is rehired. When the data warehouse wants to have a single key for that employee, a new durable key must be created that is persistent and does not change in this situation. This key is sometimes referred to as a durable supernatural key. The best durable keys have a format that is independent of the original business process and thus should be simple integers assigned in sequence beginning with 1. While multiple surrogate keys may be associated with an employee over time as their profile changes, the durable key never changes.

Drilling Down

Drilling down is the most fundamental way data is analyzed by business users. Drilling down simply means adding a row header to an existing query; the new row header is a dimension attribute appended to the GROUP BY expression in an SQL query. The attribute can come from any dimension attached to the fact table in the query. Drilling down does not require the definition of predetermined hierarchies or drill-down paths.

Degenerate Dimensions

Sometimes a dimension is defined that has no content except for its primary key. For example, when an invoice has multiple line items, the line item fact rows inherit all the descriptive dimension foreign keys of the invoice, and the invoice is left with no unique content. But the invoice number remains a valid dimension key for fact tables at the line item level. This degenerate dimension is placed in the fact table with the explicit acknowledgment that there is no associated dimension table. Degenerate dimensions are most common with transaction and accumulating snapshot fact tables.

Denormalized Flattened Dimensions

In general, dimensional designers must resist the normalization urges caused by years of operational database designs and instead denormalize the many-to-one fixed depth hierarchies into separate attributes on a flattened dimension row. Dimension denormalization supports dimensional modeling’s twin objectives of simplicity and speed.

Multiple Hierarchies in Dimensions

Many dimensions contain more than one natural hierarchy. For example, calendar date dimensions may have a day to week to fiscal period hierarchy, as well as a day to month to year hierarchy. Location intensive dimensions may have multiple geographic hierarchies. In all of these cases, the separate hierarchies can gracefully coexist in the same dimension table.

Flags and Indicators as Textual Dimension Attributes

Cryptic abbreviations, true/false flags, and operational indicators should be supplemented in dimension tables with full text words that have meaning when independently viewed. Operational codes with embedded meaning within the code value should be broken down with each part of the code expanded into its own separate descriptive dimension attribute.

Null Attributes in Dimensions

Null-valued dimension attributes result when a given dimension row has not been fully populated, or when there are attributes that are not applicable to all the dimension’s rows. In both cases, we recommend substituting a descriptive string, such as Unknown or Not Applicable in place of the null value. Nulls in dimension attributes should be avoided because different databases handle grouping and constraining on nulls inconsistently.

Calendar Date Dimensions

Calendar date dimensions are attached to virtually every fact table to allow navigation of the fact table through familiar dates, months, fiscal periods, and special days on the calendar. You would never want to compute Easter in SQL, but rather want to look it up in the calendar date dimension. The calendar date dimension typically has many attributes describing characteristics such as week number, month name, fiscal period, and national holiday indicator. To facilitate partitioning, the primary key of a date dimension can be more meaningful, such as an integer representing YYYYMMDD, instead of a sequentially-assigned surrogate key. However, the date dimension table needs a special row to represent unknown or to-be-determined dates. Filtering and grouping should be based on the date dimension’s attributes, not the smart key. When further precision is needed, a separate date/time stamp can be added to the fact table. The date/time stamp is not a foreign key to a dimension table, but rather is a standalone column. If business users constrain or group on time-of-day attributes, such as day part grouping or shift number, then you would add a separate time-of-day dimension foreign key to the fact table.


Role-Playing Dimensions

A single physical dimension can be referenced multiple times in a fact table, with each reference linking to a logically distinct role for the dimension. For instance, a fact table can have several dates, each of which is represented by a foreign key to the date dimension. It is essential that each foreign key refers to a separate view of the date dimension so that the references are independent. These separate dimension views (with unique attribute column names) are called roles.

Junk Dimensions

Transactional business processes typically produce a number of miscellaneous, low-cardinality flags and indicators. Rather than making separate dimensions for each flag and attribute, you can create a single junk dimension combining them together. This dimension, frequently labeled as a transaction profile dimension in a schema, does not need to be the Cartesian product of all the attributes’ possible values, but should only contain the combination of values that actually occur in the source data.

Snowflaked Dimensions

When a hierarchical relationship in a dimension table is normalized, low-cardinality attributes appear as secondary tables connected to the base dimension table by an attribute key. When this process is repeated with all the dimension table’s hierarchies, a characteristic multilevel structure is created that is called a snowflake. Although the snowflake represents hierarchical data accurately, you should avoid snowflakes because it is difficult for business users to understand and navigate snowflakes. They can also negatively impact query performance. A flattened denormalized dimension table contains exactly the same information as a snowflaked dimension.

Outrigger Dimensions

A dimension can contain a reference to another dimension table. For instance, a bank account dimension can reference a separate dimension representing the date the account was opened. These secondary dimension references are called outrigger dimensions. Outrigger dimensions are permissible, but should be used sparingly. In most cases, the correlations between dimensions should be demoted to a fact table, where both dimensions are represented as separate foreign keys.


Integration via Conformed Dimensions

Conformed Dimensions

Dimension tables conform when attributes in separate dimension tables have the same column names and domain contents. Information from separate fact tables can be combined in a single report by using conformed dimension attributes that are associated with each fact table. When a conformed attribute is used as the row header (that is, the grouping column in the SQL query), the results from the separate fact tables can be aligned on the same rows in a drill-across report. This is the essence of integration in an enterprise DW/ BI system. Conformed dimensions, defined once in collaboration with the business’s data governance representatives, are reused across fact tables; they deliver both analytic consistency and reduced future development costs because the wheel is not repeatedly re-created

Shrunken Rollup Dimensions

Shrunken dimensions are conformed dimensions that are a subset of rows and /or columns of a base dimension. Shrunken rollup dimensions are required when constructing aggregate fact tables. They are also necessary for business processes that naturally capture data at a higher level of granularity, such as a forecast by month and brand (instead of the more atomic date and product associated with sales data). Another case of conformed dimension subsetting occurs when two dimensions are at the same level of detail, but one represents only a subset of rows.

Drilling Across

Drilling across simply means making separate queries against two or more fact tables where the row headers of each query consist of identical conformed attributes. The answer sets from the two queries are aligned by performing a sort-merge operation on the common dimension attribute row headers. BI tool vendors refer to this functionality by various names, including stitch and multipass query.

Value Chain

A value chain identifies the natural flow of an organization’s primary business processes. For example, a retailer’s value chain may consist of purchasing to ware- housing to retail sales. A general ledger value chain may consist of budgeting to commitments to payments. Operational source systems typically produce transactions or snapshots at each step of the value chain. Because each process produces unique metrics at unique time intervals with unique granularity and dimensionality, each process typically spawns at least one atomic fact table.

Enterprise Data Warehouse Bus Architecture

The enterprise data warehouse bus architecture provides an incremental approach to building the enterprise DW/BI system. This architecture decomposes the DW/ BI planning process into manageable pieces by focusing on business processes, while delivering integration via standardized conformed dimensions that are reused across processes. It provides an architectural framework, while also decomposing the program to encourage manageable agile implementations corresponding to the rows on the enterprise data warehouse bus matrix. The bus architecture is technology and database platform independent; both relational and OLAP dimensional structures can participate.

 Enterprise Data Warehouse Bus Matrix

The enterprise data warehouse bus matrix is the essential tool for designing and communicating the enterprise data warehouse bus architecture. The rows of the matrix are business processes and the columns are dimensions. The shaded cells of the matrix indicate whether a dimension is associated with a given business process. The design team scans each row to test whether a candidate dimension is well-defined for the business process and also scans each column to see where a dimension should be conformed across multiple business processes. Besides the technical design considerations, the bus matrix is used as input to prioritize DW/BI projects with business management as teams should implement one row of the matrix at a time.

The detailed implementation bus matrix is a more granular bus matrix where each business process row has been expanded to show specific fact tables or OLAP cubes. At this level of detail, the precise grain statement and list of facts can be documented.

Opportunity/Stakeholder Matrix

After the enterprise data warehouse bus matrix rows have been identified, you can draft a different matrix by replacing the dimension columns with business functions, such as marketing, sales, and finance, and then shading the matrix cells to indicate which business functions are interested in which business process rows. The opportunity/stakeholder matrix helps identify which business groups should be invited to the collaborative design sessions for each process-centric

Continua nos próximos Posts.

Até lá!!!smurf

2 Respostas to “Kimball Dimensional Modeling Techniques – Part II”

  1. Claude Says:

    It’s hard to find your posts in google. I found it on 13 spot, you should build quality backlinks , it
    will help you to rank to google top 10. I know how to help you, just type
    in google – k2 seo tricks

  2. Durable Surrogate Key Oracle | Surrogacy Chat Says:

    […] Kimball Dimensional Modeling Techniques – Part II | … – Ralph Kimball introduced the data warehouse/business intelligence industry to dimensional modeling in 1996 with his seminal book, The Data Warehouse Toolkit…. […]

Deixe uma resposta

Preencha os seus dados abaixo ou clique em um ícone para log in:

Logotipo do

Você está comentando utilizando sua conta Sair / Alterar )

Imagem do Twitter

Você está comentando utilizando sua conta Twitter. Sair / Alterar )

Foto do Facebook

Você está comentando utilizando sua conta Facebook. Sair / Alterar )

Foto do Google+

Você está comentando utilizando sua conta Google+. Sair / Alterar )

Conectando a %s

%d blogueiros gostam disto: