Web posed new challenges: big data, cloud
grid, multi-tenancy, high availability, social media connectivity, and others. In
solving the challenges, new types of databases have been developed and become
popular. Even though they all go under the umbrella of NoSQL, each one was
developed for and is best at solving specific problems. Martin Fowler[1] argues that the future of
data storage in the enterprise is in polyglot persistence. “Polyglot persistence will occur over the
enterprise as different applications use different data storage technologies.”
My only point of dispute is that the same was true before NoSQL. The below diagram shows the two most recent waves of the database innovation: data warehouses, and Web x.0.
 |
| Waves of Database Innovation |
Applications
used before and continue to use the most suitable data storage mechanism. Data is
flowing through an enterprise and takes different forms: an in-memory storage for
a web application, multidimensional cubes for interactive reporting, bi-temporal
for master data management, flattened result sets for statistical analysis, hierarchical
XML messages for interfaces, and so on.
With the same data taking different forms, it needs to be transformed between those forms. Some representations miss constraints imposed by others and cannot be converted directly. A relational schema played the intermediary role for many years. In my opinion, it is still the best one for expressing business domains, and would continue to be used for systems of record.
With the same data taking different forms, it needs to be transformed between those forms. Some representations miss constraints imposed by others and cannot be converted directly. A relational schema played the intermediary role for many years. In my opinion, it is still the best one for expressing business domains, and would continue to be used for systems of record.
[1]
Martin Fowler and Pramod Sadalage, February 8, 2012. http://martinfowler.com/articles/nosql-intro.pdf
No comments:
Post a Comment