The increased popularity of analytical databases in recent years is mainly due to the limitations of current OLTP relational and multidimensional databases as the technology foundation for implementing a data warehouse.
For a long time, the data warehouse primarily supported tactical decision support. Numbers of users and data volumes were kept within controllable limits, and the data was usually kept up-to-date during operation. Changes were rare.
In recent years, however, the demands of companies have steadily increased on both a tactical and operational level.
The desire for faster query performance for various kind of workloads, for more complex and advanced analysis methods as well as more frequent data updates (latency reduction) has put pressure on existing data warehouse architectures. They cannot support business departments with the flexible and efficient data management they now require.
Instead of static report queries, business departments now expect more agility in terms of data provision, data processing and data preparation from IT, and increased flexibility when it comes to analytics (i.e. self-service BI). They need to react as quickly as possible to changes and new requirements.
Since this has not been the case in most organizations to date, business departments are increasingly taking steps to develop and operate their own analytical systems locally, or at the very least they are calling for more commitment from their IT departments.
However, a rethinking of the existing data warehouse architecture has also become necessary to set out the prerequisites for smart and flexible data management driven mainly by new and emerging technical requirements.
Above all, the use of “big data” involves numerous technical, conceptual and organizational changes. A central requirement in this context is to be able to carry out explorative analyses from structured and polystructured data sources – beyond the quality-assured data space that IT provides.
And another factor has increased the pressure on IT and BI managers and brought analytical databases into the debate: the high system costs of existing data warehouse environments.
To address the problems and requirements of historically grown data warehouse environments described above, analytical databases appear to constitute a relevant building block in the solution: they provide functionality for both classic BI and big data demands, promise good maintainability and can offer high performance at a lower cost (maybe in conjunction with additional software components like Hadoop or Spark).
They are becoming more affordable due to falling hardware prices and, with their high processing speed, these technologies can make the complex data architectures of existing data warehouse databases leaner, thereby adding flexibility to data management overall.