Costin Raiciu, Gianni Antichi
NDP is a datacenter network architecture and stack that was awarded best paper at Sigcomm 2017; it is also one of the first works that explicitly aims at providing both low latency and high through- put simultaneously. We describe the progress of the field since the publication of this article.
David Waledn
Brief notes relating to the Computer Communication Review in 1975–1976.
Wesley Chu
ACM SIGCOMM has reached its fiftieth birthday. The field is still remarkably strong, expanding into new disciplines as well as new application areas. Since I was an early SIGCOMM chair (1973-1977), I have been asked to recall the early days and discuss possible future directions.
Jeffrey C. Mogul, Christopher A Kantarjiev
We look back at our 1987 paper, “Fragmentation Considered Harmful,” to explain why we wrote it, how the prevalence of fragmentation was reduced by approaches such as Path MTU Discovery, and how fragmentation-related issues still lurk in today’s Internet. Our paper listed several reasons why we thought fragmentation was harmful; some were more true in 1987 than they are today, and after our paper was published, the community realized that fragmentation (and the mechanisms used to mitigate it) exposed harms we did not anticipate in our paper.
Jeffrey C. Mogul, Christopher A Kantarjiev
The original Ethernet design used CSMA/CD on a broadcast cable. Even after it became commercially popular, many people expressed concerns that Ethernet could not efficiently use the full channel bandwidth. In our 1988 paper, “Measured Capacity of an Ethernet: Myths and Reality,” we reported on experiments we ran showing that, even under relatively heavy loads, Ethernet typically still performed well. We describe the context in which we ran those experiments, and some subsequent research conducted by others.
Bruce Davie
On the occasion of the fiftieth anniversary of SIGCOMM, this article contains the reflections of a past SIGCOMM chair on the field of networking over the past few decades.
Christophe Diot
When I embarked on the CCR adventure 15 years ago I did not expect it to be so exciting, fruitful, and life changing. I am describing here our motivation and approach, our successes and failures, with I hope a perceptible sense of humor.
Craig Partridge
A brief description of how SIGCOMM’s quarterly newsletter evolved into an entry-level journal.
D
This July 2019 issue contains two techni- cal papers and three editorial notes. In ”Securing Linux with a Faster and Scalable IPtables”, Sebastiano Miano and his colleagues revisit how Linux firewalls work. Since version 2.4.0 of the Linux kernel, iptables has been the standard way of defining firewall rules in Linux. These iptables are widely used, but writing and maintaining them can be difficult. Furthermore, they have some limitations in terms of performance. This paper leverages the eBPF virtual machine that is included in the Linux kernel to propose a replacement for iptables that preserves their semantics while providing im- proved performance. They release their implementation and evaluate its performance in details.
In ”Towards Passive Analysis of Anycast in Global Routing: Unintended Impact of Remote Peering”, Rui Bian et al. analyse the deployment of anycast services. For this, they rely on different BGP routing information and highlight the impact of remote peering on anycast performance. They release their data and analysis scripts.
In addition to these two peer-reviewed papers, this issue contains three editorials. In ”Privacy Trading in the Surveillance Capitalism Age: Viewpoints on ‘Privacy- Preserving’ Societal Value Creation”, Ranjan Pal and Jon Crowcroft reconsider the current Mobile App ecosystem from an economical and privacy viewpoint. They show that the current model is not the only possible one and propose the idea of a regulated privacy trading mechanism that provides a better compromise between privacy and the commercial interests of companies. In ”Datacenter Congestion Control: Identifying what is essential and making it practical”, Aisha Mushtaq et al. take a step back at the datacenter congestion control problem. They argue that congestion control mechanisms that use Shortest-Remaining- Processing-Time are the best solution and discuss in the paper how commodity switches could be modified to support it. Finally, in ”The 11th Workshop on Active Internet Measurements (AIMS-11) Workshop Report”, kc Claffy and Dave Clark summarise the discussions at the latest AIMS workshop. They mention several new measurement ini- tiatives and interesting research projects.
Sebastiano Miano, Matteo Bertrone, Fulvio Risso, Mauricio Vásquez Bernal,
Yunsong Lu, Jianwen Pi
The sheer increase in network speed and the massive deployment of containerized applications in a Linux server has led to the consciousness that iptables, the current de-facto firewall in Linux, may not be able to cope with the current requirements particularly in terms of scalability in the number of rules. This paper presents an eBPF-based firewall, bpf-iptables, which emulates the iptables filtering semantic while guaranteeing higher throughput. We compare our implementation against the current version of iptables and other Linux firewalls, showing how it achieves a notable boost in terms of performance particularly when a high number of rules is involved. This result is achieved without requiring custom kernels or additional software frameworks (e.g., DPDK) that could not be allowed in some scenarios such as public data-centers.