The Case For In-Network Computing On Demand
The Case For In-Network Computing On Demand
复制标题
DOI：
10.1145/3302424.3303979
复制
发表时间：
2019-03
期刊：
影响因子：
--
通讯作者：
Yuta Tokusashi;Huynh Tu Dang;F. Pedone;R. Soulé;Noa Zilberman
中科院分区：
文献类型：
--
作者：
Yuta Tokusashi;Huynh Tu Dang;F. Pedone;R. Soulé;Noa Zilberman
可编程网络硬件可以运行传统上部署在服务器上的服务，从而实现性能的数量级提高。然而，尽管性能有了这些改进，网络运营商仍然对网内计算持怀疑态度。传统观点认为，功耗增加带来的运营成本超过了任何性能优势。除非网络内计算能够证明其成本合理，否则它将被视为另一项学术活动而被忽视。在本文中，我们通过对几个网络内计算用例进行详细的功耗分析来挑战这一假设。我们的实验表明，网络内计算可以非常节能。事实上，对于一瓦特，使用 FPGA 可以将商用 CPU 上的软件系统提高 100 倍，使用 ASIC 实现可以提高 1000 倍。然而，这种效率取决于系统负载。为了应对不断变化的工作负载，我们提出按需网络内计算，其中服务可以在服务器和网络之间动态移动。通过按需改变服务的布局，数据中心可以优化性能和能效。
Programmable network hardware can run services traditionally deployed on servers, resulting in orders-of-magnitude improvements in performance. Yet, despite these performance improvements, network operators remain skeptical of in-network computing. The conventional wisdom is that the operational costs from increased power consumption outweigh any performance benefits. Unless in-network computing can justify its costs, it will be disregarded as yet another academic exercise. In this paper, we challenge that assumption, by providing a detailed power analysis of several in-network computing use cases. Our experiments show that in-network computing can be extremely power-efficient. In fact, for a single watt, a software system on commodity CPU can be improved by a factor of x100 using an FPGA, and a factor of x1000 utilizing ASIC implementations. However, this efficiency depends on the system load. To address changing workloads, we propose in-network computing on demand, where services can be dynamically moved between servers and the network. By shifting the placement of services on-demand, data centers can optimize for both performance and power efficiency.