This article follows the analysis of the strengths and weaknesses of RDBMSs . He goes back to two other factors of choosing a relational database : application support and purchase criteria.
Application Support Issues
When evaluating a DBMS , do not forget to ask yourself this question: how will the databases support your applications? Are these applications with traditional transactions or batch processes? Or do you develop 2.0 applications? Do you support IoT?
The majority of traditional data management and analysis applications, including transactional processing (OLTP) and batch processing, mixed workloads, and business intelligence, are well suited to use databases. relational data.
Other specific situations encourage the use of the RDBMS data model: if the definitions and structure of the data are consistent; when the integrity and accuracy of the data must be immediate; and for processing traditional data types such as numbers, dates, and alphanumeric values.
It is advisable to adopt a relational DBMS as a standard and to deviate from it only for projects that do not take advantage of the current relational functionalities, for example Web 2.0 projects, streaming data and analytics. big data whose patterns are not fixed.
Criteria for purchasing an RDBMS
When purchasing a database system, regardless of its type, different standard criteria must be taken into account during the tender procedure and the evaluation period.
First, the architecture of the DBMS and its suitability for projects. The RDBMS architecture is suitable for most data management needs, but may be problematic in projects involving flexible schemas or complex relationships between data elements.
Also consider the availability and robustness of database administration features . Thus, backup and restore, change and performance management are essential administrative functions. RDBMS products are well equipped in this regard. There are also many complementary modules.
The deployment , including installation conditions the hardware and software prerequisites and virtualization capabilities, is an aspect not to be overlooked either. The majority of RDBMS offers robust procedures and functionality in each of these areas.
Quote: “It’s easier to outsource skills in relational databases than in other emerging technologies”
Be sure to also consider the availability of qualified personnel in selecting your DBMS. Evaluate the availability and skills of database administrators and application developers; take into account years of experience and technical certifications. It is easier to outsource skills in relational databases than in other emerging technologies.
To measure the potential effectiveness of a DBMS in relation to your needs, one of the most important elements is probably benchmarking . However, the collection of relevant performance information is far from easy.
Of course, you will find standard comparative evaluations with the Transaction Processing Performance Council . But such a testbed is quickly exceeded: it is rarely a reliable indicator of the actual performance of an implementation at the end customer.
Quote: “Some vendors, like Oracle, prohibit customers from publishing information about the performance of their DBMSs”
In addition, the contractual terms of some vendors, including Oracle, prohibit customers from publishing performance information for their database applications. The best way to do this is to review the published test benches and ask the publishers to give you reference clients.
Another option is to install a test RDBMS and develop dummy applications to evaluate the performance of the DBMS with your data, but the approach is time-consuming.
Investment in a DBMS is long-term: scalability to support the growth of data, users and processes is therefore an important factor. Support for this growth typically involves the ability to distribute data across the different nodes of a distributed system . This scalability has other implications: you must understand how the product adapts to hardware upgrades and understand the limitations of its architecture. Of course, RDBMS products offer good evolution capabilities. However, for very large volumes of data, other DBMS solutions may be better suited.
Finally, be sure to check the fault tolerance of each DBMS. A DBMS must be able to support logical and coding errors without collapsing.
In addition, a database management system relies on various components to provide its data processing services. A fault-tolerant DBMS must therefore continue to operate, even at a reduced level, rather than stopping completely in the event of a failure of one of its components or a component that it uses.