OLAP Architecture and types

OLAP
• Online Analytical Processing (OLAP) is a technology that helps businesses
analyze large amounts of data to make better decisions. OLAP is used in
business intelligence (BI) and is a core part of many data warehouses.
• How it Works
OLAP combines data from multiple sources into an OLAP cube. The cube
organizes data by dimensions, such as customers, time period, and
geographic region. OLAP cubes are often pre-summarized across
dimensions to speed up query time.
OLAP is a key part of business intelligence, which also includes relational
databases, report writing, and data mining.
 Characteristics of OLAP
• Multidimensional Data Representation: OLAP systems organize data in a
multidimensional format, where different dimensions (such as time, geography,
products, and more) are interconnected. This allows users to analyze data from
various perspectives, enabling deeper insights.
• Cubes and Hypercubes: OLAP systems use structures called “cubes” or
“hypercubes” to store and represent data. These cubes contain aggregated and
pre-calculated values for different combinations of dimensions, facilitating quick
query response times.
• Dimension Hierarchies: Dimensions within OLAP cubes often have hierarchies.
For instance, a time dimension might have hierarchies like year > quarter > month
> day. These hierarchies allow users to navigate and analyze data at different
levels of granularity.
• Fast Query Performance: OLAP systems are optimized for quick query
performance. Aggregated data stored in cubes, along with indexing and pre-
calculations, enables rapid response times for analytical queries, even over large
datasets.
• Drill-down and Roll-Up: Users can “drill down” to view more detailed data or
“roll up” to see higher-level summaries. This capability to navigate between
different levels of granularity aids in exploring data relationships.
• Slicing and Dicing: “Slicing” involves selecting a specific value along one dimension
to view a cross-section of data. “Dicing” involves selecting specific values along
multiple dimensions. These operations allow users to focus on specific subsets of
data.
• Advanced Calculations: OLAP systems support various calculations beyond simple
aggregation, such as ratios, percentages, and moving averages. These calculations
aid in deriving meaningful insights from the data.
• User-Friendly Interface: OLAP systems typically come with user-friendly interfaces
that facilitate intuitive navigation and exploration of data. This makes it easier for
non-technical users to perform complex analyses.
• Business Intelligence Integration: OLAP is often integrated with business
intelligence (BI) tools and reporting platforms. This integration allows users to
create interactive dashboards, reports, and visualizations based on OLAP data.
• Ad-Hoc Queries: Users can create ad-hoc queries to answer specific analytical
questions without needing to follow a predetermined query path. This flexibility is
crucial for exploring unexpected insights.
 Types of OLAP
• MOLAP (Multidimensional OLAP): MOLAP (Multidimensional OLAP)
systems store data in a multidimensional cube structure, with aggregated
data based on several dimensions contained in each cell of the cube.
MOLAP systems do precalculations and store aggregations, which results
in quick query responses. They work effectively in situations when
performance is crucial and data quantities aren’t very huge. Microsoft
Analysis Services, IBM Cognos TM1, and Essbase are a few MOLAP
system examples.
• Relational OLAP (ROLAP): Traditional relational databases are used for
data storage by ROLAP systems. They run intricate SQL queries to
simulate multidimensional views of the data. ROLAP systems can manage
huge datasets and complicated data linkages, therefore they can have
slightly slower query speed than MOLAP systems, but they also provide
better flexibility and scalability. ROLAP systems include those from
Oracle OLAP, SAP BW (Business Warehouse), and Pentaho, as examples.
• Hybrid OLAP (HOLAP): HOLAP systems attempt to combine the benefits of
MOLAP and ROLAP. Similar to MOLAP, they enable the ability to obtain detailed
data from the underlying relational database as necessary while also storing
summary data in cubes. Depending on the type of analysis, this method helps to
improve both performance and flexibility. Users of some MOLAP systems have
the option of retrieving detailed data or pre-aggregated data by using HOLAP
capabilities that are supported by these systems.
• DOLAP (Desktop OLAP): Desktop OLAP, often known as DOLAP, is a
simplified form of OLAP that operates on individual desktop PCs. It is appropriate
for lone analysts who wish to carry out fundamental data exploration and analysis
without requiring a large IT infrastructure. In-memory processing is frequently
used by DOLAP tools to deliver comparatively quick performance on tiny
datasets. The PivotTable feature in Excel is an illustration of a DOLAP tool.
• WOLAP (Web OLAP): WOLAP systems bring OLAP capabilities to web
browsers, allowing users to access and analyze data through a web-based interface.
This enables remote access, collaboration, and sharing of analytical insights.
WOLAP systems often use a combination of MOLAP, ROLAP, or HOLAP
architectures on the backend. Web-based BI tools like Tableau, Power BI, and
Looker provide WOLAP features.
 OLAP Architecture

• There are two main architectural approaches

in OLAP: Multidimensional (MOLAP) and
Relational (ROLAP). Here’s an explanation of
both:
 Multidimensional OLAP (MOLAP) Architecture
• MOLAP architecture is designed around the concept of multidimensional cubes.
These cubes store pre-aggregated data based on various dimensions, enabling fast
query response times. The architecture involves the following components:
• Cubes: The core element of MOLAP, cubes are multidimensional structures that
store data in cells at intersections of dimensions. Each cell contains aggregated
data or measures.
• Dimensions: Dimensions are the various perspectives or attributes by which data
can be analyzed. Common examples include time, geography, product categories,
etc. Dimensions are organized in hierarchies that allow users to drill down or roll
up for more detailed or summarized views.
• Measures: Measures are the data values that are being analyzed, such as sales
revenue, profit, quantities sold, etc. Measures are stored in the cells of the cube
and can be aggregated across dimensions.
• Aggregations: MOLAP systems pre-calculate aggregations during data processing
and store them in the cube. This speeds up query response times because
calculations are performed in advance.
• Calculation Engine: The calculation engine of the MOLAP system handles
complex calculations, aggregations, and formula-based operations on the
data.
• Storage: Data in MOLAP is stored in proprietary formats optimized for fast
retrieval. This storage structure contributes to the quick query
performance of MOLAP systems.
• User Interface: MOLAP systems provide user-friendly interfaces that allow
users to interact with multidimensional data, performing operations like
slicing, dicing, drilling, and pivoting.
Relational OLAP (ROLAP) Architecture
ROLAP architecture utilizes relational databases as the backend for data
storage and processing. It involves the following components:
• Relational Database: Data is stored in relational tables within a database.
Each table contains dimensions, measures, and the relationships between
them.
• Metadata: Metadata describes the relationships between dimensions,
measures, and other elements. It’s used to generate SQL queries that
retrieve and combine data for analysis.
• Dimension Tables: These tables store the attributes and hierarchies of
dimensions. Each row in a dimension table represents a unique dimension
value.
• Fact Tables: Fact tables store the measures and foreign keys that connect
to dimension tables. Fact tables contain the data that is being analyzed.
• SQL Engine: ROLAP systems use SQL queries to retrieve data from the
relational database based on user requests. These queries can involve
complex joins and calculations.
• Aggregations (Optional): Similar to MOLAP, ROLAP systems can employ
pre-calculated aggregations to improve query speed.
• User Interface: Users can construct and execute SQL queries, see results,
and produce reports using the interfaces provided by ROLAP systems.