Mullins Consulting's Craig Mullins discussed the current crop of DBMSs, how organizations are using them, and how they factor into the changing requirements of DBAs at Data Summit Connect Fall 2020.
Videos of full presentations from Data Summit Connect Fall 2020, a 3-day series of data management and analytics webinars presented by DBTA and Big Data Quarterly, are also now available for on-demand viewing on the DBTA YouTube channel.
Looking at the current landscape for database management systems, Mullins said there are a variety of relational and the newer element of NoSQL systems, which are being deployed to support specific use cases, and specific types of workloads. According to Mullins, this has given rise to this term "polyglot persistence," which simply means using the right database platform for each specific requirement, rather than trying to force fit everything into a single DBMS. Each DBMS has specific benefits, pros and cons for how and why they may be used, said Mullins.
Another strong trend in database management systems is database systems is a combination of multiple models in a single DBMS, such as document and key/value or graph and relational. DBAs, said Mullins, need to understand the use cases for each so that they can guide their organizations to choose appropriately. This is different than the prevailing relational-only mindset of just a few years ago when every new project used one of the major DBMS systems.
And, even without these new types of database systems, most organizations used more than one relational DBMS before NoSQL came on the scene. An organization might have an Oracle, SQL Server, MySQL, PostgreSQL, and so on, and on top of that, there are still organizations, usually larger organizations, running mainframe systems that use pre-relational databases, such as IMS or IDMS.
And then there's this additional trend that Gartner refers to as hybrid transactional analytical processing (HTAP). "And really what this means is that we're using a single DBMS to deliver both transaction processing and analytics workloads." A lot of times, organizations split that out, running their transactional workload on a relational database and their analytical workload on some kind of a warehouse appliance, said Mullins, noting that increasingly, these workloads are being brought together under a single model with the hybrid transactional analytical processing model.
Beyond those, another clear trend is that more work is being done in memory, with platforms such as SAP HANA. And then there are technologies such as Hadoop and Spark that can be used for long-term data persistence needs when the schema may not be known up-front or the schema needs to be flexible for uses such as data lakes.
"The bottom line here is that organizations on average have between 100 and 500 database instances, and those run across multiple platforms and multiple products. Really, the net net of that all is that things are getting more and more heterogeneous. And at the same time more complex."
To watch Mullins' full Data Summit Connect Fall 2020 presentation, go to the DBTA YouTube channel.