Just about everyone agrees these days that there is more data to manage than ever. The topic on which nobody seems to agree is who needs to manage it, or to what degree it might actually need to be managed.
Traditionally, database administrators (also known as DBAs) have been at the center of the data-management universe: There was always a need to have someone optimize the performance of applications by making sure data was well structured. But with the rise of Hadoop and other Big Data platforms, there’s no longer a premium on structure. In fact, many programmers are choosing to write their applications to Hadoop or other classes of so-called “NoSQL databases” to specifically eliminate the need to rely on having a DBA.
The end result has not only been a significant diminishing of the influence of the average DBA; the location where SQL queries within a data warehousing application are processed is beginning to shift.
Most organizations adopted Hadoop, at first, as a comparatively inexpensive way to store massive amounts of raw data that would be subsequently fed into a traditional data warehouse, where an analytics application could make use of it. But multiple types of SQL engines now run natively on top of Hadoop, which has profound implications for DBAs—because when it comes to managing Hadoop, there isn’t a place for a traditional administrator.
“There’s no such thing as optimizing Hadoop in the sense there is using a relational database,” said Robin Bloor, chief analyst with The Bloor Group, who believes that Hadoop-centric administrators face an altogether different set of tasks than the DBAs of old.
So how much demand will there be for classic DBAs going forward? There will always be transaction-processing applications invoking structured data—but even there, the rise of NoSQL alternatives such as Apache Cassandra is changing the way processing is done.
While some would argue that the whole notion of having a DBA could soon be archaic, others believe the role will evolve. Instead of having DBAs, the industry will see the emergence of more data architects that have skills that span multiple types of data management systems.
Part of those skill sets is making sure the integrity of the data is maintained. “When it comes to maintaining the integrity of the data the DBA still has a lot of those skills,” said Steve Palmer, senior vice president of global data and analytics for Avanade, an IT service firm that is a joint venture between Accenture and Microsoft.
In addition to data-management needs, the location of Big Data warehousing applications in the age of the cloud is also likely to change. Increasingly, data is analyzed where it is created. Rather than moving massive amounts of data across wide area networks, a more hybrid approach to data warehousing is emerging, wherein analytics applications run wherever the data was originally created. This “Data Gravity” approach to managing data will challenge administrators, who will be required to manage federated data warehouses running both on public and private clouds.
In the end, the people responsible for managing all those data warehouses may no longer be DBAs in the classic sense of the relational database term. They may wind up, for example, being known as data architects. But whatever the name of the function, most of the tasks associated with managing data—at least as enterprise IT organizations have always known them—are not going away anytime soon.
- Do You Need an Oracle Database Specialization?
- The Key Skills Needed by Big Data Engineers
- Despite Apps, Much of Data Analytics Still Drudgery
Image: Arjuna Kodisinghe/Shutterstock.com