Dynamic Data Routing Decisions for Compliant Data Handling in Service-and Cloud-Based Architectures: A Performance Analysis

Abstract

In many service-based applications, decisions about data routing need to be made at runtime, for instance to ensure compliant data handling. Different service-and cloudbased architectures to make dynamic data routing decisions exist including central entities, multiple dedicated dynamic router services, or using a sidecar for each involved service. These architectures differ in various quality attributes including complexity, understandability, and changeability of the decision logic. Choosing the wrong architecture for decision-making at runtime may severely impact the performance of the software system. In this paper, we have evaluated the performance of three representative approaches for processing compliance rules concerned with data routing in service-and cloud-based architectures. The results show that distributed approaches for dynamic data routing have a better performance compared to centralized solutions. On the other hand, centralized solutions are easier to understand and change, but this strongly depends on the domain problem.

Modeling and Empirical Validation of Reliability and Performance Trade-Offs of Dynamic Routing in Service- and Cloud-Based Architectures

“…In our prior work [2], [3], we have studied representative service-and cloud-based system architecture patterns for dynamic request routing. A typical cloud-native architecture pattern is the Sidecar pattern [18], [24] in which the sidecar of each service handles incoming and outgoing traffic [11].…”

Automatic Adaptation of Reliability and Performance Trade-Offs in Service- and Cloud-Based Dynamic Routing Architectures

“…In our prior work [3]- [5], we studied three representative dynamic routing architectures (see Figure 1). We created an analytical model of request loss as a metric of system reliability, and performed a multiple regression analysis to statistically model performance using round-trip times.…”

“…In our prior work [3]- [5], we have studied three representative service-and cloud-based dynamic routing architecture patterns: Central Entity based architectures (CE), as shown in Figure 1a, e.g., an API Gateway [29] or any kind of central service bus [12]; Sidecar Architectures (SA), as presented in Figure 1b, which follows the sidecar pattern [15], [22], [25], and Dynamic Routers architectures (DR) [19] in which multiple routers perform the routing for groups of services as shown in Figure 1c. DR can be seen as a hybrid of the central entity and the sidecar architectures; SA is a completely distributed approach since there is a sidecar per each service.…”

“…pattern [22], [29]. In our prior work [3]- [5], we empirically studied these architecture patterns in terms of system reliability and performance. Our studies show that more centralized routing results in a higher reliability; however, decentralized routing offers a higher performance.…”

Cost-Aware Multidimensional Auto-Scaling of Service- and Cloud-Based Dynamic Routing to Prevent System Overload

“…In our prior work [2]- [5], we studied three representative dynamic routing architecture patterns in service-and cloudbased environments. These include the central-entity-based architecture, e.g., an API Gateway [18], or any kind of central service bus [8]; dynamic routers architecture [12] in which multiple routers are responsible for the routing regarding groups of services; and the sidecar architecture that follows the sidecar pattern [13].…”