The Current State of Database Technology (VIDEO)

Video produced by Steve Nathans-Kelly

There are more options than ever for data management, including relational, columnar, and in-memory. In a Data Summit presentation, titled "Designing a Data Architecture for Modern Business Intelligence & Analytics," Athena IT solutions managing partner Richard Sherman offered an overview of today's database technologies and their roles.

"I'll give you the high-level synopsis of where we are on databases," he said. "We still continue to have the relational database as a key toward transactional processing and data warehousing and master data management. Although, OLAP cubes are certainly used for BI, the best practice at this point is columnar databases, whether it's wide column databases, columnar databases, or column stores within Oracle or SQL Server; one of those combinations, that's generally the best practice for BI."

DBTA’s next Data Summit conference will be held May 19-20, 2020, in Boston, with pre-conference workshops on Monday, May 18.

Hadoop has also been used heavily, and as well as—in the NoSQL space, key-value, column, graph, and column databases, wide column databases. "All of those have specific use cases, all of these could be used within the analytical data architecture, based on use cases."

In addition, there are in-memory analytics, in-database analytics, MPP databases, in-memory, cloud, DB appliances—all those are really rolling into the idea of a multi-model, multi-schema, or polygon database. So in the long run, you'll get a database management system that will have relational, columnar, and other constructs built within them, Hadoop, and it will manage where the data goes or you'll maybe coax it a little bit, but you'll have your data spread out across different data structures based on use cases.

"Relational doesn't solve everything, but neither does Hadoop, so you need the combination of what works best for you," he observed.

"Some of the top vendors are AWS, Azure, Google—mainly on machine learning and column databases—and then although Oracle's a distant fourth, as long as Larry [Ellison] breathes, I wouldn't count them out as far as being a key player in the marketplace. IBM and others—they're not exhibiting here, are they? They're not so much. I have lots of customers with Oracle, they have lots of great stuff, my nephew works for IBM. But those are the top, top ones and they'll evolve, and again mainly the advantage that Microsoft has now with Azure is that it has a hybrid architecture, so you can have stuff on-premise and in the cloud. Not everybody is going to be in the cloud right away, or maybe ever. That's really a cost issue and other issues related to it."

All of these vendors are providing different types of database structures, so that organizations can put their data in the space that makes the most sense, said Sherman. "And again, data warehousing, MDM, relational, hybrid dimensional models, columnar would be a logical dimensional model, and it could be either a wide column, a column in a database like Vertica, or column stores, the column indexes that are put into SQL Server, Oracle, and other databases themselves. All of them do the same equivalent thing, and some are more efficient than others. And, data virtualization has a combination of a bunch of these things embedded in the tools too."

Many presenters have made their slide decks available on the Data Summit 2019 website at www.dbta.com/DataSummit/2019/Presentations.aspx.