A Comprehensive Study on Optimizing Systems with Data Processing Units

Abstract

New hardware, such as SmartNICs, has been released to offload network applications in data centers. Off-path Smart-NICs, a type of multi-core SoC SmartNICs, have attracted the attention of many researchers. Unfortunatelly, they lack the fully exploration of off-path SmartNICs.In this paper, we use a BlueField SmartNIC as an example to conduct a systematical study on the advantages and disadvantages of off-path SmartNICs. We make a detailed performance characterization on an off-path SmartNIC including computing power and network communication overhead, and propose the following advices: 1) Directly utilize the specific accelerators on the SmartNIC to offload applications; 2) Offload latency-insensitive background processing to the Smart-NIC to reduce the load on the host; 3) Regard the SmartNIC as a new endpoint in the network to expand the computing power and storage resources of the server host; 4) Avoid directly employing the design method for systems based on on-path SmartNICs. We apply these advices to several use cases and show the performance improvements.

Efficient traffic monitoring on SmartNICs through XDP co-design

“…During the communication process between VM1 and VM2, only the first packet needs to be processed by the host. Other data packets are offloaded through the OVS data plane to the SmartNIC, they all bypass the kernel, thereby achieving the purpose of releasing valuable CPU resources on the host side and accelerating data packet forwarding [5,6].…”

“…While OVS itself provides some commands to collect the statistics of the offloading flows, it often misses the information on the mice flow offloaded since these flows are very short. Other methods of getting statistics from the SmartNIC inevitably compromise the performance [5,6]. The statistics provided by the SmartNIC are limited by the provider, which is very inflexible and cannot be obtained on demand.…”

“…Smart network cards emerged as the times required and gradually became the "third core" of the data center by providing various acceleration and hardware offloading capabilities, releasing expensive CPU resources, and providing better scalability and higher performance, which is widely used [1][2][3][4][5][6].…”