Base De Datos

Introduction
ETL is an acronym for Extract, Transform and Load – the three types of activity required to move data from one or
more source systems to one or more receiving systems. This is outlined in the diagram below. These activities are
explained in more detail later in this paper. ETL is normally thought of as being one of the components of a
complete Business Intelligence solution -with Data Warehouse representing the storage component, Visualisation
and Analysis Tools representing the applications which deliver the actual business benefit, and ETL providing the
bridge between the Business Intelligence systems and the transactional systems which generate the raw source
data. But this is only a part of the truth. There are many circumstances in which organisations need tomanage the
movement of data between systems in order to reflect changes in one system to other related ones.
ETL has always been a requirement since the very early days of IT, but initially the number of systems and the
complexity of data was sufficiently low that hand-coded bespoke applications were the predominant solution. In the
current IT environment it is no longer possible to build andmaintain the number and complexity of applications
required.
ETL products are available from a number of ETL specialist vendors. In addition several larger vendors – in
particular database vendors – have recently started to bundle ETL components with their products in order to build
a more attractive and more competitive package. The products from specialist vendors still deliver worthwhileperformance and functionality benefits over the bundled products, but users must evaluate their own requirements
to determine the best fit. This paper aims to highlight some of the issues.
The last section of this paper identifies a number of products which have been evaluated by Bloor Research and for
which evaluation papers are available.
ETL and Business Intelligence
As Data Warehousesgrow in scope, the effort of manually coding programs to manage the movement of data from
transactional systems to data warehouse becomes difficult to cost justify. Often multiple skill sets are required,
particularly where data must be extracted from source systems which execute on a variety of proprietary platforms,
and with data stored in all manner of file systems and legacy databases.
Notonly has the scope of Data Warehouses grown, but the interest in long-term trends and increasing levels of
detail has caused an explosion of the size requirements. This brings associated challenges to both the database
technology and to the processes of controlling the content within the warehouse. This is the heart of the demand for
ETL products.
Business Intelligence has different demandson ETL compared with data movement between transactional systems.
Most organisations wish to keep a Data Warehouse which is known to be consistent at a given point in time, with
regular updates being scheduled according to business requirements. This typically means that extremely large
sets of data have to be processed in a time-restricted maintenance period.
ETL - Extract, Transform, Loadhttp://www4.bloor-research.com/00000085/00000085.html (1 of 6) [11/19/1999 3:39:14 PM]
ETL and application implementations
The applications sector of IT has been dominated for the last few years by integrated "ERP" (Enterprise Resource
Planning) packaged software. This type of package aims to provide a seamless flow of data through related
business processes, from receipt of an order, throughall the internal processes of order fulfillment, to the eventual
billing and ongoing accounting functions. More recently a further type of integrated suite "CRM" (Customer
Relationship Management) addresses another business sector. The ultimate aim in each case is to provide a
customiseable and comprehensive set of functions which do not suffer from the fragmentation of earlier 'piecemeal'...