一种支持高并发的感知数据接入服务

Cloud Receiver Service for High-concurrency Sensory Data

随着物联网技术的广泛应用,体现物理世界运行状态的感知数据被大量前端传感设备采集;为了在云端有效存储、管理及后续分析这些数据,迫切需要支持海量前端设备接入不同种类、持续不断的异构数据.当前相关技术和系统在接入大规模高并发数时存在吞吐量和执行效率的瓶颈,为此设计实现了一种支持高并发的感知数据接入服务.该服务面向异构感知数据的通信,通过可定制数据结构和异步I/O多路复用,实现了大规模异构感知数据的高并发接收、解析和分发处理.相关的实验表明,该服务即使在低配虚拟机中也可以支持万级并发的数据接入能力.

In Interet of Things, massive data can be gathered from physical sensors. In order to store, manage and analyze sensory data effectively,receiver service is required to collect huge data from heterogeneous ,continuous and multi-type sources. Current technologies and systems have the throughput and efficiency bottleneck when in high concurrency conditions ;in this paper we propose a Cloud receiver service to collect heterogeneous sensory data from multiple sources based on a customized tuple structure with asynchronous I/O. In extensive experiments,our service can receive sensory data smoothly from ten-thousand connections even on a low virtual machine.