Roll-Up or Drill-Down, Which has the largest Data Warehouse Detail?

1

Context

Studying data warehouse (DW) I came across content regarding the types of OLAP servers and the types of operations that can be done, within that context I was confused about the concept of detailing two operations, they are called Roll-Up and Drill-Down

In the book the author states that the Drill-Down operation has greater detailing but the given example did not make me well understood.

Exemplification of the Book.

Chapter 29 - Data Warehousing and OLAP Overview. Here's how operations are visually exemplified by the author of the book.

  

Roll-Up Example

     
    

  
  

Drill-DownExample

    
    

  

WhattheTeradatapointWebsiteAffirms

Astheexplanationwasnotveryclear,Iwenttogoogleandfoundthis teradatapoint article that made me understand better.

  

About Roll-Up Operation

     
    

  

My Understanding

The content on the site teradatapoint, made me better understand the subject and if I could summarize or explain in a few words the operations, it would look something like this:

  

Roll-Up Operation seeks to reduce the 3D cube for a leaner analysis, for example: Perform a bimetral analysis of a given product in a given region.

  

Drill-Down operation seeks to increase the 3D cube, for an analysis with a longer time interval, for example performing analyzes at annual intervals of a given product in a given region.

I understand by detailing the data, an increasingly narrow delimitation of the scope involved, which indicates decrease in cube dimension size, but the book states the opposite.

Doubt

I have not understood which of the two operations there is a greater detailing of the data , the book states that it is Drill-Down and the terapoint website says Roll-Up. >

Source: Database Systems 6th edition, authors: Elmasri, Ramez Navathe, Shamkant B. Year: 2011 Cap 29 pg 724 and 725

    
asked by anonymous 17.11.2017 / 16:50

1 answer

1

I'm new here, and this is my first answer. Since my knowledge of the subject is not one of the deepest, I will try to be as brief as possible.

OLAP

When the database is built on OLAP, it means that it is no longer relational, ie there is no increase or decrease in the amount of data, they are already in a format whose recovery is usually faster than in a database of relational data. However, there are versions of OLAP, ranging from ROLAP (with a relational engine), MOLAP (with a multidimensional engine) and a HOLAP (Hybrid) intermediary. Unfortunately I have never worked with any hybrids to give further explanations.

Drill-Down and Roll-Up

When you access information from an OLAP database (that is, Multidimensional), it may have two or more dimensions. Some of these dimensions may have hierarchies. The higher hierarchies are the aggregate information of the lower hierarchies, for example: One dimension side can be constructed with customer location information. In this case, Brazil would be the highest hierarchy. When applying a Drill-down, this hierarchy can be divided in several ways, for example regions (North, South, Southeast etc.) or states (SP, RJ etc.). The lower you go down the detail information, the more chance you have of finding unmeasured cells in the same way that the information becomes detailed.

As for Roll-Up, one can think of the product dimension as it becomes easier to understand. You might be organizing the database of a retail store with multiple products. Therefore each product would be the lowest cell in the hierarchy of this dimension. Thus, in applying the aggregation of the sales of a certain group of products (Liquifier -> Kitchen Products -> Appliances, for example), the complication in this point is the fact that the information is not standardized among the products, for example, which will require some form of data transformation.

I hope I have answered. Hugs!

    
21.01.2018 / 00:03