Analytic Workloads: Which Data Warehouse Architecture Is Right for You?
The number of workloads for analytics has increased significantly in recent years. This is because more user organizations than ever before are using advanced forms of analytics, including complex SQL, data mining, and statistical analysis. Also, new forms of analytics have recently gained a foothold, including natural language processing, data visualization, and Hadoop MapReduce.
One thing in common across these analytic workloads is that most are demanding and unpredictable when executed. And most have data preparation requirements that differ from those of other workloads hosted by an enterprise data warehouse (EDW). Hence, user organizations that are diving deeper into analytics are asking: Do we execute analytic workloads in the EDW proper? Or should they go onto secondary analytic platforms that integrate with the EDW on some level? Depending on how you answer these questions, you may need to re-architect and beef up your EDW and/or acquire additional analytic platforms.
You Will Learn:
- The analytic workloads that are available today, plus what each one does
- How the requirements of analytic workloads differ from those commonly hosted by an EDW, namely reporting, performance management, and OLAP
- How the rise of analytics is driving a trend toward distributed data warehouse architectures and designs
- Tips for choosing platforms and architectures for analytic workloads
Philip Russom, Ph.D.