Distributed System Data Architectures

Data Architecture in Distributed Systems

• The logic level distributed architecture relates to the conceptual schema of the database, while the physical level relates to the internal schema. These architectures implement aids and/or limitations on functionality or access rights.

Traditional Data Model Classification

Centralized Data

The data are organized in a single database that is accessed by the entire distributed system. This was the first model to appear, as the need to design client/server systems arose where HOST data was organized in this way.

Partitioning Model

The data are distributed across multiple databases spread across the distributed platform.

Replicated Data

Copies of "stable" data from centralized databases are created and mounted on a centralized model "close" to clients.

• There are three ways of replicating data:
  • Copy: Data is copied exactly as it is.
  • Summarize: Data is aggregated to a higher level.
  • Rearrange: The original data model is converted to eliminate unnecessary attributes and/or combine entities.

Multibase

• The integrated database maintains each individual logical schema. Programs are used to resolve semantic and syntactic differences.
• The logic for shared entities is encapsulated in servers. Programs allow clients or other servers using them to interact as if with a single entity.

Distributed Database

• The integrated database has a single conceptual framework, allowing individual databases (BDs) to be viewed as one logical level. Each integrated database corresponds to a physical level. The designer must know which databases are remote to monitor performance and response times.
• It is generally better to work with a distributed database than with a multibase system.

Federated Database

• It is a special case of a distributed database involving different, independent owners. A common conceptual framework is agreed upon, often resolved through views.
• Each database must link to the federation and accept the common understanding.

OLAP (Online Analytical Processing)

• Also known as multidimensional databases, OLAP allows users to quickly analyze information organized into multidimensional views and hierarchies.
• For example, OLAP tools perform trend analysis on sales and financial data, enabling users to delve into large volumes of statistics to identify trends or volatile products.

Multidimensional OLAP (MOLAP)

• Traditional OLAP products that summarize transactions into multidimensional views beforehand, resulting in very fast user queries.

Relational OLAP (ROLAP)

• This term applies to tools that extract data and create multidimensional views using complex SQL statements on traditional relational databases. They can create pseudo-OLAP queries on the fly.