How to Measure the Killer Microsecond

PrimoracMia,; BugnionEdouard,; ArgyrakiKaterina,

doi:10.1145/3155055.3155065

Cited by 6 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Middlebox machine runs one of the four NFs mentioned above (we use one core). The Tester machine runs MoonGen [22] to generate traffic and measure packet loss, throughput, and latency; for the latency measurements, we rely on hardware timestamps for better accuracy [49]. We use DPDK v. 16.07 on Ubuntu Linux with kernel 3.13.0-119-generic.…”

Section: Performance Evaluationmentioning

confidence: 99%

A Formally Verified NAT

Zaostrovnykh

Pirelli

Pedrosa

et al. 2017

Proceedings of the Conference of the ACM Special Interest Group on Data Communication

Self Cite

View full text Add to dashboard Cite

We present a Network Address Translator (NAT) written in C and proven to be semantically correct according to RFC 3022, as well as crash-free and memory-safe. There exists a lot of recent work on network verification, but it mostly assumes models of network functions and proves properties specific to network configuration, such as reachability and absence of loops. Our proof applies directly to the C code of a network function, and it demonstrates the absence of implementation bugs. Prior work argued that this is not feasible (i.e., that verifying a real, stateful network function written in C does not scale) but we demonstrate otherwise: NAT is one of the most popular network functions and maintains per-flow state that needs to be properly updated and expired, which is a typical source of verification challenges. We tackle the scalability challenge with a new combination of symbolic execution and proof checking using separation logic; this combination matches well the typical structure of a network function. We then demonstrate that formally proven correctness in this case does not come at the cost of performance. The NAT code, proof toolchain, and proofs are available at [58].

show abstract

Section: Performance Evaluationmentioning

confidence: 99%

A Formally Verified NAT

Zaostrovnykh

Pirelli

Pedrosa

et al. 2017

Proceedings of the Conference of the ACM Special Interest Group on Data Communication

Self Cite

View full text Add to dashboard Cite

show abstract

“…Dedicated timestamping hardware [35,36] offers line-rate high-precision and high-accuracy timestamping on multiple 10G Ethernet ports but requires additional hardware, increasing the costs of the measurement setup. A study by Primorac et al [37] compared MoonGen's timestamping to various software and hardware timestamping solutions. They concluded that MoonGen's hardware timestamping method offers a similar accuracy and precision compared to a professional timestamping hardware solution.…”

Section: Background and Related Workmentioning

confidence: 99%

“…Switching off energysaving mechanisms can improve latency beyond the 99.99th percentile by approx. 10 s according to Primorac et al [37].…”

Section: Os-specific Techniquesmentioning

confidence: 99%

How Low Can You Go? A Limbo Dance for Low-Latency Network Functions

et al. 2022

View full text Add to dashboard Cite

Throughput is a commonly used performance indicator for networks. However, throughput may be considered insignificant if data is outdated or networks become unpredictable or unreliable. Critical services may even prioritize latency, predictability, and reliability at the expense of throughput to avoid detrimental effects on service operation. Latency, predictability, and reliability are distinct qualities realized in real-time systems. Real-time systems often require additional effort using non-standard interfaces, requiring customized software, or providing low throughput figures. This work picks up the challenge and investigates a single-server network function—a building block for end-to-end low-latency network applications. Assessing reliability and quantifying low latency is equally challenging, as sub-microsecond latency and $$1/10^{5}$$ 1 / 10 5 loss probability leave little room for error. Both, our measurement and the investigated platforms, rely on Linux running on off-the-shelf components. Our paper provides a comprehensive study on the impact of various components on latency and reliability, such as the central processing unit (CPU), the Linux Kernel, the network card, virtualization features, and the networking application itself. We chose Suricata, an intrusion prevention system (IPS), representing a widely deployed, typical network application as our primary subject of investigation.

show abstract

Seiðr: Dataplane Assisted Flow Classification Using ML

Simpson

Cziva

Pezaros

2020

GLOBECOM 2020 - 2020 IEEE Global Communications Conference

View full text Add to dashboard Cite

Real-time, high-speed flow classification is fundamental for network operation tasks, including reactive and proactive traffic engineering, anomaly detection and security enhancement. Existing flow classification solutions, however, do not allow operators to classify traffic based on fine-grained, temporal dynamics due to imprecise timing, often rely on sampled data, or only work with low traffic volumes and rates. In this paper, we present Seiðr, a classification solution that: (i) uses precision timing, (ii) has the ability to examine every packet on the network, (iii) classifies very high traffic volumes with high precision. To achieve this, Seiðr exploits the data aggregation and timestamping functionality of programmable dataplanes. As a concrete example, we present how Seiðr can be used together with Machine Learning algorithms (such as CNN, k-NN) to provide accurate, real-time and high-speed TCP congestion control classification, separating TCP BBR from its predecessors with over 88-96 % accuracy and F1-score of 0.864-0.965, while only using 15.5 MiB of memory in the dataplane.

show abstract

How to Measure the Killer Microsecond

Cited by 6 publications

References 15 publications

A Formally Verified NAT

A Formally Verified NAT

How Low Can You Go? A Limbo Dance for Low-Latency Network Functions

Seiðr: Dataplane Assisted Flow Classification Using ML

Contact Info

Product

Resources

About