Grus: Enabling Latency SLOs for GPU-Accelerated NFV Systems

Zheng, Zhilong; Bi, Jun; Wang, Haiping; Sun, Chen; Yu, Heng; Hu, Hongxin; Gao, Kai; Wu, Jianping

doi:10.1109/icnp.2018.00025

Cited by 19 publications

(8 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Modern virtualized data centers usually have servers running the network access layer [2,110], using a software switch that connects virtual machines to the physical network [19,144,160,193]. Driven by these requirements, over the past years, software-based packet processing has made significant inroads in the traditionally hardware-dominated network appliance market [54,73,162] with several established programmable software switch platforms for efficient network virtualization (VPP [19], BESS [77], FastClick [20], NetBricks [157], PacketShader [78], and ESwitch [144]), user space I/O libraries (PacketShader [8], NetMap [169], Intel DPDK [87], RDMA [98], FD.io [56], and Linux XDP with eBPF [22]), and NFV platforms [101,112,192,196,207].…”

Section: General-purpose Hardwarementioning

confidence: 99%

The Programmable Data Plane

Michel

Bifulco²,

Rétvári

et al. 2021

ACM Comput. Surv.

View full text Add to dashboard Cite

Programmable data plane technologies enable the systematic reconfiguration of the low-level processing steps applied to network packets and are key drivers toward realizing the next generation of network services and applications. This survey presents recent trends and issues in the design and implementation of programmable network devices, focusing on prominent abstractions, architectures, algorithms, and applications proposed, debated, and realized over the past years. We elaborate on the trends that led to the emergence of this technology and highlight the most important pointers from the literature, casting different taxonomies for the field, and identifying avenues for future research.

show abstract

Section: General-purpose Hardwarementioning

confidence: 99%

The Programmable Data Plane

Michel

Bifulco²,

Rétvári

et al. 2021

ACM Comput. Surv.

View full text Add to dashboard Cite

show abstract

“…Modern virtualized data centers usually have servers running the network access layer [2,110], using a software switch that connects virtual machines to the physical network [20,143,158,191]. Driven by these requirements, over the past years, software-based packet processing has made significant inroads in the traditionally hardware-dominated network appliance market [55,74,160] with several established programmable software switch platforms for efficient network virtualization (VPP [20], BESS [78], FastClick [21], NetBricks [155], PacketShader [79], and ESwitch [143]), user space I/O libraries (PacketShader [8], NetMap [167], Intel DPDK [88], RDMA [99], FD.io [57], and Linux XDP with eBPF [23]), and NFV platforms [102,112,190,194,205].…”

Section: General-purpose Hardwarementioning

confidence: 99%

The Programmable Data Plane: Abstractions, Architectures, Algorithms, and Applications

Michel¹,

Bifulco²,

Rétvári³

et al. 2020

Preprint

View full text Add to dashboard Cite

<div><div>Programmable data plane technology enables the systematic reconfiguration of the low-level processing steps applied to network packets and is a key driver in realizing the next generation of network services and applications. This survey presents recent trends and issues in the design and implementation of programmable network devices, focusing on prominent architectures, abstractions, algorithms, and applications proposed, debated, and realized over the past years. We elaborate on the trends that led to the emergence of this technology and highlight the most important pointers from the literature, casting different taxonomies for the field and identifying avenues for future research.</div></div>

show abstract

“…Lemur targets macro-level SLOs across NF chains on many platforms, and can use ResQ's techniques. Grus [47] lowers latency for NFs on GPUs given a bound; in future work we can extend Lemur to GPUs using techniques from Grus.…”

Section: Related Workmentioning

confidence: 99%

Meeting SLOs in cross-platform NFV

Yen

Wang

Supittayapornpong

et al. 2020

Proceedings of the 16th International Conference on Emerging Networking EXperiments and Technologies

View full text Add to dashboard Cite

Network Functions (NFs) perform on-path processing of network traffic. ISPs are deploying NF Virtualization (NFV) with software NFs run on commodity servers. ISPs aim to ensure that NF chains, directed acyclic graphs of NFs, do not violate Service Level Objectives (SLOs) promised by the ISP to its customers. To meet SLOs, NFV systems sometimes leverage on-path hardware (such as programmable switches and smart NICs) to accelerate NF execution. Lemur places and executes NF chains across heterogeneous hardware while meeting SLOs. Lemur's novel placement algorithm yields an SLO-satisfying NF placement while weighing many constraints: hardware memory and processing stages, server cores, link capacity, NF profiles, and NF chain interactions. Lemur's metacompiler automatically generates code and rules (in P4, Python, eBPF, C++, and OpenFlow) to stitch cross-platform NF chain execution while also optimizing resource usage. Our experiments show that Lemur is alone among competing strategies in meeting SLOs for canonical NF chains while maximizing marginal throughput (the traffic rate in excess of the service-level objective).

show abstract

Grus: Enabling Latency SLOs for GPU-Accelerated NFV Systems

Cited by 19 publications

References 22 publications

The Programmable Data Plane

The Programmable Data Plane

The Programmable Data Plane: Abstractions, Architectures, Algorithms, and Applications

Meeting SLOs in cross-platform NFV

Contact Info

Product

Resources

About