difference between database and a data warehouse

Upload: vangavasu

Post on 07-Apr-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/6/2019 Difference Between Database and a Data Warehouse

    1/1

    Q: What is a difference between database and a data warehouse? Why can't weuse database as a data warehouse?

    A: Roman Bukary, SAP director of market strategy: Today a data warehouse isbuilt on top of a database. A data warehouse is a specialized format of data st

    orage designed to place all relevant business data within a series of PK-FK (primary-foreign key) connected tables for easy querying and analysis. These tablesstore granular data, as well as some aggregates and indices for rapid response,

    in this specialized format to enable easy "slice and dice" across multiple represented dimensions so that you could ask and get answers to your questions such as: what is the total volume of all SKUs sold last month, what is the most popular SKU in a give retail outlet, how do various SKUs contribute to your overall profitability. To create and manage PK-FK, table joins (to answer queries such aswho are our top customers who buy the most per week), updates, data integrity (row-level locking), baseline security, data administration and so on, data warehouses rely on underlying databases. So, in fact, you use a database and build a data warehouse on top of it just as you would use a database and build a CRM or financial application on top.Don Hatcher, vice president of technology strategy for SAS: A data warehouse isone usage of database technology. The difference is in the design of the data mo

    del.Operational databases are designed to collect and update vast amounts of information from many users, and collect it a little at a time, in many short transactions. They are primarily data collectors and are optimized for data collection.Data warehouses are databases that are designed for just the opposite usage. They are designed for relatively few complex transactions (queries), executed infrequently and seek a large result set that can be surfaced to relatively few users. They are primarily data distributors and are optimized for data distribution.In some cases, you can use the same database engine (DBMS) for both usages, although our research has proven that databases designed specifically for data warehousing will outperform those designed for data collection, even after tuning those transactional databases for data warehouse performance.Allen Houpt, director of product management for CleverPath solutions at CA: The

    notion of database in the context is an online transaction processing (OLTP) application. The main reason for loading transactional data into a data warehouse is the need to report information over time. To keep the high efficiency that isrequired for transactional systems, there is a limited time range that is kept in the database. Often, also, there is little descriptive information that is kept in the OLTP database that makes the data less understandable to business users. Basically OLTP are based on events (point in time), whereas the data warehouseis based on information over time.