In today\'s fast-paced business environment, we see an ongoing trend towards the need for analytics on the latest operational data. The data management layer of enterprise applications needs to adapt to this requirement and In-Memory Column Stores have been proposed as a new architecture that can handle such mixed workload scenarios. A thorough understanding of the resulting query workload is required to validate and optimize data management concepts for this new challenge. Consequently, this paper introduces Database Application Context (DAC) analysis - an holistic framework to analyze database workloads, data characteristics as well as access patterns on specific domain types. We present results for a productive enterprise resource planning system, as well as widely accepted database benchmarks for transactional and mixed workloads. In contrast to existing work, we have analyzed correlations between issued queries and the domain types of accessed attributes. Our main findings are (i) that enterprise workloads are read heavy, (ii) that specific database operators predominantly operate on attributes with a specific domain type, and (iii) that data characteristics differ depending on the data type. Furthermore, based on our analysis of trends in modern enterprise applications, we expect workloads with an increased runtime share of complex queries in the future. These findings help in designing and optimizing the data management layer of modern enterprise applications.
|Cite as: Wust, J., Meyer, C. and Plattner, H. (2014). DAC: Database Application Context Analysis applied to Enterprise Applications. In Proc. Thirty-Seventh Australasian Computer Science Conference (ACSC 2014) Auckland, New Zealand. CRPIT, 147. Thomas, B. and Parry, D. Eds., ACS. 39-48 |
(local if available)