Answer / suba

multidimensional online analytical processing and
relational online analytical processing. In MOLAP data is
stored in form of multidimensional cubes. The advantages of
this mode is that it provides excellent query performance
and the cubes are built for fast data retrieval. All
calculations are pre-generated when the cube is created and
can be easily applied while querying data.In ROLAP, the
data is stored in relational databases this model gives the
appearance of traditional OLAP’s slicing and dicing
functionality. The advantages of this model is it can
handle a large amount of data and can leverage all the
functionalities of the relational database.

Answer / nethaji

MOLAP has aggregated value stored in cube.Since the data is
aggregated, query performance is fast.
ROLAP has data sored in relational databases.Here query has
to access the database for retrieving the data every time.So
performance is slow when compared to molap. Size is larger
than molap.

Answer / edar

The idea of adopting the relational technology to store
data to a data warehouse has a solid foundation if you
consider the huge amount of literature written about the
relational model, the broadly available corporate
experience with relational database usage and management,
and the top performance and flexibility standards of
relational DBMSs (RDBMSs). The expressive power of the
relational model, however, does not include the concepts of
dimension, measure, and hierarchy, so you must create
specific types of schemata so that you can represent the
multidimensional model in terms of basic relational
elements such as attributes, relations, and integrity
constraints. This task is mainly performed by the well-
known star schema. See Chapter 8 for more details on star
schemata and star schema variants.

The main problem with ROLAP implementations results from
the performance hit caused by costly join operations
between large tables. To reduce the number of joins, one of
the key concepts of ROLAP is denormalization—a conscious
breach in the third normal form oriented to performance
maximization. To minimize execution costs, the other key
word is redundancy, which is the result of the
materialization of some derived tables (views) that store
aggregate data used for typical OLAP queries.

From an architectural viewpoint, adopting ROLAP requires
specialized middleware, also called a multidimensional
engine, between relational back-end servers and front-end
components. The middleware receives OLAP queries formulated
by users in a front-end tool and turns them into SQL
instructions for a relational back-end application with the
support of meta-data. The so-called aggregate navigator is
a particularly important component in this phase. In case
of aggregate views, this component selects a view from
among all the alternatives to solve a specific query at the
minimum access cost.

In commercial products, different front-end modules, such
as OLAP, reports, and dashboards, are generally strictly
connected to a multidimensional engine. Multidimensional
engines are the main components and can be connected to any
relational server. Open source solutions have been recently
released. Their multidimensional engines (Mondrian, 2009)
are disconnected from front-end modules (JPivot, 2009). For
this reason, they can be more flexible than commercial
solutions when you have to create the architecture (Thomsen
and Pedersen, 2005). A few commercial RDBMSs natively
support features typical for multidimensional engines to
maximize query optimization and increase meta-data
reusability. For example, since its 8i version was made
available, Oracle's RDBMS gives users the opportunity to
define hierarchies and materialized views. Moreover, it
offers a navigator that can use meta-data and rewrite
queries without any need for a multidimensional engine to
be involved.

Different from a ROLAP system, a MOLAP system is based on
an ad hoc logical model that can be used to represent
multidimensional data and operations directly. The
underlying multidimensional database physically stores data
as arrays and the access to it is positional (Gaede and
Günther, 1998). Grid-files (Nievergelt et al., 1984; Whang
and Krishnamurthy, 1991), R*-trees (Beckmann et al., 1990)
and UB-trees (Markl et al., 2001) are among the techniques
used for this purpose.

The greatest advantage of MOLAP systems in comparison with
ROLAP is that multidimensional operations can be performed
in an easy, natural way with MOLAP without any need for
complex join operations. For this reason, MOLAP system
performance is excellent. However, MOLAP system
implementations have very little in common, because no
multidimensional logical model standard has yet been set.
Generally, they simply share the usage of optimization
techniques specifically designed for sparsity management.
The lack of a common standard is a problem being
progressively solved. This means that MOLAP tools are
becoming more and more successful after their limited
implementation for many years. This success is also proven
by the investments in this technology by major vendors,
such as Microsoft (Analysis Services) and Oracle (Hyperion).

The intermediate architecture type, HOLAP, aims at mixing
the advantages of both basic solutions. It takes advantage
of the standardization level and the ability to manage
large amounts of data from ROLAP implementations, and the
query speed typical of MOLAP systems. HOLAP implies that
the largest amount of data should be stored in an RDBMS to
avoid the problems caused by sparsity, and that a
multidimensional system stores only the information users
most frequently need to access. If that information is not
enough to solve queries, the system will transparently
access the part of the data managed by the relational
system. Over the last few years, important market actors
such as MicroStrategy have adopted HOLAP solutions to
improve their platform performance, joining other vendors
already using this solution, such as Business Objects.

Answer / pratap gangula

In addition to the above answers, ROLAP storage models are better and faster for real time reporting & analytics whereas MOLAP storage models takes time to update itself. ROLAP models are slower because every query fired on cube has to satisfy RDBMS policies as well.

Answer / pratap gangula

Hi Marcelo,

The retrieval procedure is different because the storage model is different so ultimately the main difference between ROLAP and MOLAP is still in the storage concepts. While designing DW we dont decide whether the requirement is to fetch the data using SQL or MDX....rather we focus on whether the need for storing the data is in RDBMS format or Multidimensional format .

Answer / pratap gangula

Hi Marcelo,

Your understanding of MDX execution flow is correct but
your interpretation of difference between ROLAP and MOLAP is
wrong. The difference between ROLAP and MOLAP lies in the way the data is stored which leads to two different data retrieval methodology.

OLAP storage models are designed basis how you want to store your data. I guess nobody would want to design the OLAP basis the data retrieval procedures.

You are simply challenging your own thought process.

Rgds,
Pratap

Answer / gopal

There are 4 OLAPS
R_OLAP
D-OLAP
M-OLAP
H-OLAP

R-OLAP- relational OLAP
EX-Cognos Impromptu,microstratagy

M-OLAP(Multidimensional OLAP)-Ex-Cubes

Answer / marcelo malheiros

Hello Pratap,

Sorry but you are missing the some steps re retrieving
information from a cube...

As you may know, a cube can have its data stored in a RDBMS
(in tables).

In order to retrieve the data from the RDBMS, when the
frontend sends a MDX instruction to the cube, the cube has
to translate the MDX instruction into SQL and send the SQL
to the RDBMS. The RDBMS returns a table as result set from
the SQL statement. The cube reads that table (result set)
and converts it to a hierarchical result set. The
hierarchical result set is then sent to the front-end that
formats the information and displays it to the end-user...

So, in the perspective of the frontend, an MDX instruction
is sent to the cube and a hierarchical result set is
received by the cube... Even if the data is in a RDBMS...

Sumarizing: Doesn't matter if the cube is self-contained or
if the information is stored elsewhere, for the frontend
the storage is transparent because a MOLAP frontend ALWAYS
sends a MDX instruction to the cube and receives a
hierarchical result set...

I hope that helps...

Cheers!
Marcelo

Answer / marcelo malheiros

I believe you lost the begining of my answer...

"I know a lot of people understand ROLAP and MOLAP as where
the information is stored: if data is in a RDBMS, is ROLAP;
if it is in a self-contained cube, it is MOLAP.

I find the above a misconcept of the terms."

Let me repeat the fundamental question: How would classify
On Line Analytical Process performed on top of IMS, IDMS,
ADABAS, VSAM, flat file, XML and other non RDBMS data
storages?"

Answer / marcelo malheiros

I know a lot of people understand ROLAP and MOLAP as where
the information is stored: if data is in a RDBMS, is ROLAP;
if it is in a self-contained cube, it is MOLAP.

What I find the above a misconcept of the terms.

I believe ROLAP and MOLAP differs from each other in the
language you use to retrieve the result set and the result
set format...

I don't find that storage is relevant... How would classify
BI done on top of IMS, IDMS, ADABAS, VSAM and other non
RDBMS data storages?... Yes, BI started using those data
repositories... As well, how would you classify BI deployed
on top of a flat file or an XML file?...

SO MY POINT IS THAT THE DATA STORAGE DOES NOT DEFINE ROLAP
OR MOLAP...

By MOLAP:
I understand that the data is retrieved via MDX and the
result set is formated as a hiearchy with tags (something
similar to XML) which provides seemless hierarchy
manipulation capabilities to MOLAP.

By ROLAP:
I understand that the data is retrieved via SQL (relational
algebra) and the result set is always formatted as a table.

My humble 2 cents...

Cheers!
Marcelo

What is the difference between star and snowflake schemas?

0 Answers  

How do you call a module named �testmod.cfm� with the parameters param1=�yes� and param2=5?

0 Answers  

What is the difference between datawarehouse and bi?

0 Answers  

What is confirmed dimension?

7 Answers   JPMorgan Chase,

What are various warehousing tools?

0 Answers  

Do you view contain data?

0 Answers  

How to join 3 database like flat file, oracle, db2 in informatrica?

0 Answers  

What is a data profile?

0 Answers  

Explain what are non-additive facts in detail?

0 Answers  

Do you always need to copy the data before reporting on it?(Example, real-time & on-demand reporting is a requirement)?

0 Answers  

How would you declare an inline css to format the table with a background color of yellow and give the table cell a right margin of 10 pixels?

0 Answers  

Explain piconet?

0 Answers