Edge computing relies on devices at the edge of the network to best meet the application’s needs, such as low communication latency and data caching. However, unbalanced networks and resource heterogeneity are challenging factors when optimizing data transfers and applications’ performances in the edge. In this paper, we explore these two concerns through a comprehensive set of benchmarks and illustrate their importance with the help of two elements that should be part of any edge computing toolkit: data locality and context awareness. Thanks to the benchmarks, we highlight some pitfalls and opportunities that may be met when trying to deploy applications on the edge.

边缘计算依靠网络边缘的设备来最好地满足应用程序的需求，例如低通信延迟和数据缓存。但是，当优化边缘的数据传输和应用程序性能时，不平衡的网络和资源异构性是具有挑战性的因素。在本文中，我们通过一套全面的基准测试来探讨这两个问题，并借助两个应作为任何边缘计算工具套件组成部分的元素来说明它们的重要性：数据局部性和上下文感知。感谢基准测试，我们着重介绍了尝试在边缘上部署应用程序时可能遇到的一些陷阱和机遇。