Data Warehouse:
The term Data Warehouse was coined by Bill Inmon in 1990, which he defined in the following way: "A warehouse is a subject-oriented, integrated, time-variant and non-volatile collection of data in support of management's decision making process". He defined the terms in the sentence as follows:
Subject Oriented:
Data that gives information about a particular subject instead of about a company's ongoing operations.
Integrated:
Data that is gathered into the data warehouse from a variety of sources and merged into a coherent whole.
Time-variant:
All data in the data warehouse is identified with a particular time period.
Non-volatile
Data is stable in a data warehouse. More data is added but data is never removed. This enables management to gain a consistent picture of the business.
(Source: "What is a Data Warehouse?" W.H. Inmon, Prism, Volume 1, Number 1, 1995).
This definition remains reasonably accurate almost ten years later. However, a single-subject data warehouse is typically referred to as a data mart, while data warehouses are generally enterprise in scope. Also, data warehouses can be volatile. Due to the large amount of storage required for a data warehouse, (multi-terabyte data warehouses are not uncommon), only a certain number of periods of history are kept in the warehouse. For instance, if three years of data are decided on and loaded into the warehouse, every month the oldest month will be "rolled off" the database, and the newest month added.
Ralph Kimball provided a much simpler definition of a data warehouse. As stated in his book, "The Data Warehouse Toolkit", on page 310, a data warehouse is "a copy of transaction data specifically structured for query and analysis". This definition provides less insight and depth than Mr. Inmon's, but is no less accurate.
Data warehousing is essentially what you need to do in order to create a data warehouse, and what you do with it. It is the process of creating, populating, and then querying a data warehouse and can involve a number of discrete technologies such as:
Source System Identification:
Source System Identification: In order to build the data warehouse, the appropriate data must be located. Typically, this will involve both the current OLTP (On-Line Transaction Processing) system where the "day-to-day" information about the business resides, and historical data for prior periods, which may be contained in some form of "legacy" system. Often these legacy systems are not relational databases, so much effort is required to extract the appropriate data.
Data Warehouse Design and Creation:
This describes the process of designing the warehouse, with care taken to ensure that the design supports the types of queries the warehouse will be used for. This is an involved effort that requires both an understanding of the database schema to be created, and a great deal of interaction with the user community. The design is often an iterative process and it must be modified a number of times before the model can be stabilized. Great care must be taken at this stage, because once the model is populated with large amounts of data, some of which may be very difficult to recreate, the model can not easily be changed.
Data Acquisition:
This is the process of moving company data from the source systems into the warehouse. It is often the most time-consuming and costly effort in the data warehousing project, and is performed with software products known as ETL (Extract/Transform/Load) tools. There are currently over 50 ETL tools on the market. The data acquisition phase can cost millions of dollars and take months or even years to complete. Data acquisition is then an ongoing, scheduled process, which is executed to keep the warehouse current to a pre-determined period in time, (i.e. the warehouse is refreshed monthly).
Changed Data Capture:
The periodic update of the warehouse from the transactional system(s) is complicated by the difficulty of identifying which records in the source have changed since the last update. This effort is referred to as "changed data capture". Changed data capture is a field of endeavor in itself, and many products are on the market to address it. Some of the technologies that are used in this area are Replication servers, Publish/Subscribe, Triggers and Stored Procedures, and Database Log Analysis.
Data Cleansing:
This is typically performed in conjunction with data acquisition (it can be part of the "T" in "ETL"). A data warehouse that contains incorrect data is not only useless, but also very dangerous. The whole idea behind a data warehouse is to enable decision-making. If a high level decision is made based on incorrect data in the warehouse, the company could suffer severe consequences, or even complete failure. Data cleansing is a complicated process that validates and, if necessary, corrects the data before it is inserted into the warehouse. For example, the company could have three "Customer Name" entries in its various source systems, one entered as "IBM", one as "I.B.M.", and one as "International Business Machines". Obviously, these are all the same customer. Someone in the organization must make a decision as to which is correct, and then the data cleansing tool will change the others to match the rule. This process is also referred to as "data scrubbing" or "data quality assurance". It can be an extremely complex process, especially if some of the warehouse inputs are from older mainframe file systems (commonly referred to as "flat files" or "sequential files").
Data Aggregation:
is process is often performed during the "T" phase of ETL, if it is performed at all. Data warehouses can be designed to store data at the detail level (each individual transaction), at some aggregate level (summary data), or a combination of both. The advantage of summarized data is that typical queries against the warehouse run faster. The disadvantage is that information, which may be needed to answer a query, is lost during aggregation. The tradeoff must be carefully weighed, because the decision can not be undone without rebuilding and repopulating the warehouse. The safest decision is to build the warehouse with a high level of detail, but the cost in storage can be extreme.
Now that the warehouse has been built and populated, it becomes possible to extract meaningful information from it that will provide a competitive advantage and a return on investment. This is done with tools that fall within the general rubric of "Business Intelligence".
Business Intelligence (BI)
A very broad field indeed, it contains technologies such as Decision Support Systems (DSS), Executive Information Systems (EIS), On-Line Analytical Processing (OLAP), Relational OLAP (ROLAP), Multi-Dimensional OLAP (MOLAP), Hybrid OLAP (HOLAP, a combination of MOLAP and ROLAP), and more. BI can be broken down into four broad fields:
Multi-dimensional Analysis Tools:
Tools that allow the user to look at the data from a number of different "angles". These tools often use a multi-dimensional database referred to as a "cube".
Query tools:
Tools that allow the user to issue SQL (Structured Query Language) queries against the warehouse and get a result set back.
Data Mining Tools:
Tools that automatically search for patterns in data. These tools are usually driven by complex statistical formulas. The easiest way to distinguish data mining from the various forms of OLAP is that OLAP can only answer questions you know to ask, data mining answers questions you didn't necessarily know to ask.
Data Visualization Tools:
Tools that show graphical representations of data, including complex three-dimensional data pictures. The theory is that the user can "see" trends more effectively in this manner than when looking at complex statistical graphs. Some vendors are making progress in this area using the Virtual Reality Modeling Language (VRML).
Metadata Management
Throughout the entire process of identifying, acquiring, and querying the data, metadata management takes place. Metadata is defined as "data about data". An example is a column in a table. The datatype (for instance a string or integer) of the column is one piece of metadata. The name of the column is another. The actual value in the column for a particular row is not metadata - it is data. Metadata is stored in a Metadata Repository and provides extremely useful information to all of the tools mentioned previously. Metadata management has developed into an exacting science that can provide huge returns to an organization. It can assist companies in analyzing the impact of changes to database tables, tracking owners of individual data elements ("data stewards"), and much more. It is also required to build the warehouse, since the ETL tool needs to know the metadata attributes of the sources and targets in order to "map" the data properly. The BI tools need the metadata for similar reasons.
Summary:
Data Warehousing is a complex field, with many vendors vying for market awareness. The complexity of the technology and the interactions between the various tools, and the high price points for the products require companies to perform careful technology evaluation before embarking on a warehousing project. However, the potential for enormous returns on investment and competitive advantage make data warehousing difficult to ignore.
Bill Inmon
Bill Inmon is universally recognized as the "father of the data warehouse." He has over 26 years of database technology management experience and data warehouse design expertise, and has published 36 books and more than 350 articles in major computer journals. His books have been translated into nine languages. He is known globally for his seminars on developing data warehouses and has been a keynote speaker for every major computing association. Before founding Pine Cone Systems, Bill was a co-founder of Prism Solutions, Inc.
Ralph Kimball
Ralph Kimball was co-inventor of the Xerox Star workstation, the first commercial product to use mice, icons, and windows. He was vice president of applications at Metaphor Computer Systems, and founder and CEO of Red Brick Systems. He has a Ph.D. from Stanford in electrical engineering, specializing in man-machine systems. Ralph is a leading proponent of the dimensional approach to designing large data warehouses. He currently teaches data warehousing design skills to IT groups, and helps selected clients with specific data warehouse designs. Ralph is a columnist for Intelligent Enterprise magazine and has a relationship with Sagent Technology, Inc., a data warehouse tool vendor. His book "The Data Warehouse Toolkit" is widely recognized as the seminal work on the subject
Data Warehousing Concepts:
Dimensional Data Model:
Dimensional data model is most often used in data warehousing systems. This is different from the 3rd normal form, commonly used for transactional (OLTP) type systems. As you can imagine, the same data would then be stored differently in a dimensional model than in a 3rd normal form model.
To understand dimensional data modeling, let's define some of the terms commonly used in this type of modeling:
Dimension: A category of information. For example, the time dimension.
Attribute: A unique level within a dimension. For example, Month is an attribute in the Time Dimension.
Hierarchy: The specification of levels that represents relationship between different attributes within a dimension. For example, one possible hierarchy in the Time dimension is Year → Quarter → Month → Day.
Fact Table: A fact table is a table that contains the measures of interest. For example, sales amount would be such a measure. This measure is stored in the fact table with the appropriate granularity. For example, it can be sales amount by store by day. In this case, the fact table would contain three columns: A date column, a store column, and a sales amount column.
Lookup Table: The lookup table provides the detailed information about the attributes. For example, the lookup table for the Quarter attribute would include a list of all of the quarters available in the data warehouse. Each row (each quarter) may have several fields, one for the unique ID that identifies the quarter, and one or more additional fields that specifies how that particular quarter is represented on a report (for example, first quarter of 2001 may be represented as "Q1 2001" or "2001 Q1").
A dimensional model includes fact tables and lookup tables. Fact tables connect to one or more lookup tables, but fact tables do not have direct relationships to one another. Dimensions and hierarchies are represented by lookup tables. Attributes are the non-key columns in the lookup tables.
In designing data models for data warehouses / data marts, the most commonly used schema types are Star Schema and Snowflake Schema.
Star Schema: In the star schema design, a single object (the fact table) sits in the middle and is radially connected to other surrounding objects (dimension lookup tables) like a star. A star schema can be simple or complex. A simple star consists of one fact table; a complex star can have more than one fact table.
Snowflake Schema: The snowflake schema is an extension of the star schema, where each point of the star explodes into more points. The main advantage of the snowflake schema is the improvement in query performance due to minimized disk storage requirements and joining smaller lookup tables. The main disadvantage of the snowflake schema is the additional maintenance efforts needed due to the increase number of lookup tables
No comments:
Post a Comment