Differentiated Real-Time Data Services for E-Commerce Applications
Electronic Commerce Research
| Hi-index | 0.00 |
Efficient real-time data management has become increasingly important as real-time applications become more sophisticated and data-intensive. In data-intensive real-time applications, e.g., online stock trading, agile manufacturing, sensor data fusion, and telecommunication network management, it is essential to execute transactions within their deadlines using fresh (temporally consistent) sensor data, which reflect the current real-world states. However, it is very challenging to meet this fundamental requirement due to potentially time-varying workloads and data access patterns in these applications. Also, user service requests and sensor updates can compete for system resources, thereby making it difficult to achieve guaranteed real-time database services in terms of both deadline miss ratio and data freshness. In this paper, we present a novel real-time data management architecture, which includes feedback control, admission control, and flexible update schemes, to enforce the deadline miss ratio of admitted user transactions to be below a certain threshold, e.g., 1%. At the same time, we maintain data freshness even in the presence of unpredictable workloads and access patterns. In a simulation study, we illustrate the applicability of our approach by showing that stringent performance specifications can be satisfied over a wide range of workloads and access patterns. We also show that our approach can significantly improve performance compared to baselines.