| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.misc -> Re: the difference between data warehouse, datamart, and just a database
Posted and Emailed to original poster
I hope this excerpt from an extensive research project my agency did will help clarify this issue:
Yes, there are different types of structures and analysis and they require different types of information characteristics. Additionally, the analysis generally require different types of users.
No, there is not a difference. A data warehouse and a data mart can use the same tools and design methods. Each may have a similar design or schema as well as utilize the same set or types of tools.
The main difference is user requirements, size and specificity. Data warehouses have been described as subject oriented, data marts are subject-specific. A data mart contains a subset of subject oriented corporate data that is of value to a specific business unit, department, or set of users. This subset consists of historical, summarize, and possibly
detailed data captured from a line of business source system (independent data marts) or from a enterprise data warehouse (a dependent data mart). Like an data warehouse, a data mart is used for short-and long-term business planning an decision-making; unlike a data warehouse, a data mart does not provide the capability to analyze data across multiple business units of an organization.
In general, a data warehouse is a broad storehouse of data for many subject areas. A data mart is generally more specific and represents a single or specific subject area or part of the business. See figure on page 9.
According to W. H. Inmon (Father of data warehousing), a "true" data mart is simply an extract of the enterprise data warehouse to contain only data that pertains to one department of the enterprise. It may or may not follow data warehouse design rules, but will always reflect accurately the enterprise view of that department, no more and no less.
Other opinions indicate different types of data marts.
"There are two basic ways that companies create data marts. First is to capture data directly from a Online Transaction Processing (OLTP) system into the marts that need it. Second is to capture data from OLTP systems into a central data warehouse and then to feed the data marts with data from the warehouse." (SQL Server 7, Data Warehousing. Pg. 23.)
There is however, a danger in feeding data directly into each data mart separately. By sending data that has not been consolidated or planned directly to a data mart, the organization creates silo systems or what is referred in data warehouse circles as a "stovepipe" system of data marts. Beyond not accessing the same the same data from centralized warehouse, these stovepipe data marts do not integrate with a central metadata repository. They do not support shared definitions of data entities, business rules, and semantics across the organization.
The current trend is towards a central data warehouse with coordinated data marts extracting data from the data warehouse.
Kristian Rickert <krickert_at_integrationware.com> wrote:
>Data Warehouses do NOT always have to be stored in an OLAP model. OLAP
>is another term that precalculates everything in a DB first which is not
>necessarily the case with a data warehouse. Both terms are used
>interchangably these days, though. And there's a million different
>definitions for it. MOST Data Warehouses are OLAP-style. Yet not all
>OLAP cubes are Data Warehouse architectured. Data Warehousing is more
>complex than OLAP technology. The two are similiar in that they both
>involve multidimentional architecture.
>
>Kristian Rickert
>IntegrationWare, Inc.
>
>
>
>
>Jerry Gitomer wrote:
>>
>> Hi,
>>
>> This is the simplistic answer suitable for managers.
>>
>> Data warehouse - a specialized database containing historical
>> data for an entire organization. The data warehouse is
>> characterized by OLAP (On Line Analytical Processing) as opposed
>> to OLTP (On Line Transaction Processing) of a conventional
>> database.
>>
>> Datamart - a "mini" data warehouse containing historical data
>> for one department of an organization. May be further limited to
>> data related to a single function or group of functions performed
>> by a department.
>>
>> regards
>> Jerry Gitomer
>>
>> kal_at_valistech.com wrote in message
>> <379023CD.BE08DFBB_at_valistech.com>...
>> >Could someone explain briefly what the differences are
>> architectually,
>> >physically .. etc?
>> >this question came up in a meeting and I didn't know how to
>> answer it.
>> >I appreciate the help.
>> >Kal.
>> >
>> >PS please copy me when responding to the newsgroup
>> >
-----------== Posted via Newsfeeds.Com, Uncensored Usenet News ==---------- http://www.newsfeeds.com The Largest Usenet Servers in the World! ------== Over 73,000 Newsgroups - Including Dedicated Binaries Servers ==----- Received on Mon Jul 19 1999 - 14:04:58 CDT
 |
 |