Let's talk about BI Layers – Series 3 of 3
Coming to the last of the blog series on BI layers, this topic will also touch upon some popular analysis models used by data architects and analysts to develop and understand an enterprise data model.
Data preparation layer – This layer is important from a data mining perspective as it is concerned with exploring large volumes of data to determine patterns and trends of information. Preparation of data for loading into data marts and pre-calculation of values to be loaded into OLAP data repositories are usually done here.
Metadata Repository Layer – Data about data is metadata and to take this data well beyond data structure names and formats is the responsibility of this layer. This layer must be comprehensive in scope covering data flowing between the various layers and not to forget transformation and validation rules.
Warehouse Management Layer – Can be considered as a security administrator and responsible for scheduling tasks to build and maintain data in data warehouse and data mart layers.
Application Messaging Layer – Other than transporting information between various layers, this layer encompasses generation and storage of control messages and its communication to the target
Internet/Intranet Layer – Basic data communication, browser based user interfaces, TCP/IP networking are all part of this layer.
Before we wind up this blog series, I wanted to mention some popular analysis models which can help understand the data model of an EDW.
Context diagrams which help outline major business processes are well known. Swim Lane Diagrams which deconstruct business processes and Entity Relationship diagrams(ER) which depict relations between data entities play an important role in developing an enterprise data model. Understanding of data’s business purpose and context helps mitigate the risk of suboptimal data in the warehouse.
