With the rise of the Internet of Things (IoT) and Edge computing, a considerable amount of system services are moving from reliable Cloud data centers to less reliable infrastructures closer to the end-users. However, the constrained resources, unreliable communication, and varying operating conditions of IoT pose significant complexities for software vendors in testing their applications. Although several emulators exist for testing IoT systems, numerous issues can be pointed out, such as lacking support for logical dependencies and advanced fault injection capabilities, requiring manual validation of the system's behavior, or focusing on a specific platform and language. To address these limitations, we propose Frisbee: a framework for the automated testing of IoT applications. Frisbee accelerates the testing process by simplifying the spin-up of distributed virtual testbeds over Kubernetes and the description of `what-if' scenarios that integrate complex sequences of workloads and faultloads.
翻译:随着Tings(IoT)和Edge(Edge)互联网的兴起,大量系统服务正在从可靠的云型数据中心转向更接近终端用户的较不可靠的基础设施,然而,IoT的资源有限、通信不可靠和操作条件各异,给软件供应商测试其应用程序带来了很大的复杂性。虽然存在一些模拟器来测试IoT系统,但可以指出许多问题,例如缺乏对逻辑依赖性和高级错误注入能力的支持,需要人工验证系统的行为,或侧重于特定的平台和语言。为了解决这些限制,我们建议Frisbee:一个自动测试IoT应用程序的框架。 Frisbee简化了库韦尔特斯上分布的虚拟试床的旋转和描述“如果”情景,将工作量和过载的复杂序列结合起来。