Database administration remains an important skill as organizations continue to store and access much of their data using a database management system. But what it means to be a DBA has changed over the years.
At a high level, DBAs are tasked with implementing, managing, and ensuring the integrity and efficiency of database systems. Some focus on logical design, others focus on physical design. Some DBAs specialize in building systems, and others specialize in maintaining and tuning systems. There are system DBAs, application DBAs, database architects, and performance specialists, and there are specialty DBAs and general-purpose DBAs.
For more articles looking at the current state of technology and what's ahead for 2021, download the Data Sourcebook.
The primary role of database “custodian” continues to be the main emphasis of the job, but that is no longer sufficient for most organizations. The DBA is expected to take on numerous additional responsibilities. These can include writing application code, enterprise application integration, managing web services, network administration, and so on.
Some organizations have gone as far as renaming the job of DBA to be database architect, database specialist, or even database developer. Nevertheless, DBAs need to constantly adapt their skillset. Let’s take a look at how DBAs’ roles are changing.
Polyglot Persistence
Today, although relational database systems still dominate for most types of applications, NoSQL database systems are being deployed for certain non-traditional use cases that include web storefront applications, real-time analytical processing, storage of shopping cart data and user profiles, social networks, routing and dispatch systems, and location-aware systems. For these types of use cases, NoSQL database systems, such as document databases, key/value stores, wide-column stores, and graph databases, can be more effective.
The selection of the best database system to use for each type of use case is referred to as polyglot persistence. All that really means is using different database systems for different applications and use cases based upon how the database supports the needs of the application.
Increasingly, DBAs are required to manage not just structured data but also unstructured data so that it can be accessed, analyzed, and modified in an efficient way by end users. Most modern DBMSs can store and manage unstructured data such as large documents, spreadsheets, images, audio, and video. Often, the DBA will also be involved in building and administering a data lake as a storage repository for raw or refined data in native format until it needs to be accessed. Frequently, data lakes use Hadoop for data storage because of its ability to support a flexible schema, so understanding of Hadoop is yet another skill required of modern DBAs.
Multi-DBMS DBAs
Today, most DBAs are required to administer multiple different database systems. At first glance, that statement may seem to have already been covered by the previous section, but it really is a separate, although related, trend. Traditionally, DBAs were expected to be in-depth experts on one specific DBMS, for example, the Db2 DBA or the Oracle DBA. But this is no longer the case.
However, it is common now for organizations to have database products that span pre-relational, relational, and NoSQL, and, frequently, multiple DBMSs from each type, which may also be open source and proprietary. Each DBMS requires different, albeit similar, skills and knowledge to administer proficiently.
DBAs must acquire the requisite knowledge and skills to manage multiple different DBMSs, which can be quite dissimilar in their capabilities and requirements. For example, managing a hierarchic DBMS such as IBM IMS is quite different from managing a relational DBMS such as IBM Db2, even though both are IBM products. Likewise, managing Oracle requires different skills and techniques than managing MongoDB. Often, DBAs are required to master multiple DBMSs and platforms, complicating the job.
Many DBMS vendors are adapting and delivering multi-model database systems, in which one product can be used to store and access data using more than one data model. For example, many NoSQL offerings combine document and key-value stores into a single product. Similarly, RDBMS platforms have also evolved to support NoSQL data models, such as adding document and graph stores to their core relational engine. HTAP, or hybrid transaction analytical processing, is also provided by some RDBMS offerings with different storage engines supported for transactions versus analytical queries. DBAs need to understand and be able to support all of these variations of using different types and suppliers of database systems.
The Rise of DevOps
At the same time that they must know about all of these new database technologies, today’s DBAs are also more aligned with development and applications than ever before. Many organizations have deployed agile development with DBAs participating in teams along with the other developers. This fosters cooperation and communication between the folks coding the application (Dev) and the folks developing the database (Ops).