However, since DBaaS is a cloud implementation, many of the safeguards that need to be applied to cloud migrations also need to be used, Kamsky added. There is some data that organizations will always prefer to host in their own private data centers. In addition, Kamsky said, for massive, established applications, it can be a complex process to lift them to the cloud.
Data Warehousing As A Service
As DBaaS gains traction, a similar paradigm, data warehousing-as-a-service (DWaaS), is also under consideration as an option for making data available to the enterprise. “This technology is evolving rapidly with new innovations, including DataOps methodologies,” said Itamar Ankorion, SVP and managing director, enterprise data integration, at Qlik. “Companies must be aware that these modern DWaaS platforms require innovation in how data is ingested and organized for analytical processing. By adopting a completely modern data architecture based on DataOps principles and technologies of continuous data integration and data warehouse automation, companies benefit from efficient data management.”
While DWaaS is still in the relatively early stages, there is “fast growth in initial or small deployments,” said Ankorion. “We expect to see continued high growth and adoption rates within the next year and mass market adoption in the next 5 years.” The challenge, he said, is that DWaaS needs complementary technologies to facilitate efficient and continuous data ingestion as well as data warehouse automation. Selecting the right technology partners or perhaps looking deeper into more modern data frameworks such as DataOps will help to achieve success, Ankorion explained.
Enter AI
Of course, no discussion of the power of data analytics is complete without considering the implications of AI and machine learning.
“We are at a stage where a lot of progress has been made on the analytics side in terms of building machine learning models, but businesses are now looking at applying those models to fast-moving transactions data in real time,” said Kumar. “On that front, AI is one of the most significant new technologies that can take deep learning models and serve them against fast-moving time series or streaming data with less than a millisecond of response time. This enables making recommendations—and intelligently acting upon data in real time—a reality. In addition, new hardware technology such as Intel Optane now enables businesses to run extremely large datasets—petabytes—with relatively low costs, but with similar performance of data management running in-memory.”
AI as a data management and analysis enabler has recently started to gain more traction, said Redis’ Kumar. Some organizations have begun experimenting with running their AI models directly inside of the transactions that are put through a time series data model. This in turn opens up new use cases such as fraud mitigation in real time.
AI is also dramatically improving enterprise big data management capabilities. “It used to be that businesses would consider using different databases for different use cases, but as the need for a dynamic layer between applications and traditional databases increases, there are some who are looking to manipulate different data models closer to the application layer,” said Kumar. “This opens up new use cases that combine data models in real time. Examples include searching graph data, combining streams data with JSON, or running AI models on time series solutions.”
Containers, Kubernetes, And Hybrid Clouds
The trend of data environments becoming increasingly flexible and capable of being moved to where the business demands them is coming from containerization via Kubernetes, which has “blurred the line between the data center and public cloud,” said Anupam Singh, general manager of data warehouse for Cloudera. “This allows data management to burst dynamically both inside the data center and the public cloud, without the knowledge of the end user or developer.”