We report on experiences with Swift congestion control in Google datacenters. Swift targets an end-to-end delay by using AIMD control, with pacing under extreme congestion. With accurate RTT measurement and care in reasoning about delay targets, we find this design is a foundation for excellent performance when network distances are well-known. Importantly, its simplicity helps us to meet operational challenges. Delay is easy to decompose into fabric and host components to separate concerns, and effortless to deploy and maintain as a congestion signal while the datacenter evolves. In large-scale testbed experiments, Swift delivers a tail latency of <50μs for short RPCs, with near-zero packet drops, while sustaining ~100Gbps throughput per server. This is a tail of <3x the minimal latency at a load close to 100%. In production use in many different clusters, Swift achieves consistently low tail completion times for short RPCs, while providing high throughput for long RPCs. It has loss rates that are at least 10x lower than a DCTCP protocol, and handles O(10k) incasts that sharply degrade with DCTCP.

This paper presents our design and experience with a microkernel-inspired approach to host networking called Snap. Snap is a userspace networking system that supports Google's rapidly evolving needs with flexible modules that implement a range of network functions, including edge packet switching, virtualization for our cloud platform, traffic shaping policy enforcement, and a high-performance reliable messaging and RDMA-like service. Snap has been running in production for over three years, supporting the extensible communication needs of several large and critical systems.

In this paper, a selective data compression scheme is developed to combine the need for efficient data transmission and the receivers' interest in changes presented in the data by taking the advantage of the fact that receivers hold a copy of previous data. Three-step pre-processing is introduced in this paper. Firstly we separate the unchanged areas (the majority) of the image from the changed areas between successive images of the same area. Secondly, the bands which are sensitive to the changes are identified with the aid of statistical measures. Finally, a binary index image of each band is generated to indicate the two categories. Following this pre-processing, compression of the unchanged areas and changed areas is conducted separately. In this way, different compression algorithms can be applied to each case. As the majority data will be unchanged and only a subset of bands reflects the changes, high compression rate is achievable.

This paper examines the deployment of the DNS Security Extensions (DNSSEC), which adds cryptographic protection to DNS, one of the core components in the Internet infrastructure. We analyze the data collected from the initial DNSSEC deployment which started over 2 years ago, and identify three critical metrics to gauge the deployment: availability, verifiability, and validity. Our results provide the first comprehensive look at DNSSEC’s deployment and reveal a number of challenges that were not anticipated in the design but have become evident in the deployment. First, obstacles such as middle-boxes (firewalls, NATs, etc.) that exist in today’s Internet infrastructure have proven to be problematic and have resulted in unforeseen availability problems. Second, the public-key delegation system of DNSSEC has not evolved as it was hoped and it currently leaves over 97 % of DNSSEC zones isolated and unverifiable, unless some external key authentication mechanism is added. Furthermore, our results show that cryptographic verification is not equivalent to validation; a piece of verified data can still contain the wrong value. Finally, our results demonstrate the essential role of monitoring and measurement in the DNSSEC deployment. We believe that the observations and lessons from the DNSSEC deployment can provide insights into measuring future Internet-scale cryptographic systems.

The question that this panel will explore is the feasibility of specialization within software engineering degree programs, particularly at the undergraduate level. Over the years of software engineering curriculum development, we have seen good success in implementation of both undergraduate and graduate degree programs. We have also seen that some universities and students have shown an interest in specialization within these degree programs.

As higher education courses are increasingly expected to have an online presence, academic staff are faced not only with workload challenges but also with changes to the type of work in which they are involved and consequently the skills, dispositions and support that is required. Within the Faculty of Education at Queensland University of Technology, several mini-cases relating to micro-innovations in the design of video in online teaching were examined for the purpose of understanding these challenges. The investigation is significant in that it takes a broader view of the learning communities involved in design to include academic peers, technical support staff and facility managers. In considering this broader learning community, it is clear that a proactive approach is needed at the faculty level to ensure that the work of innovators and early adopters is recognised and shared in a way that contributes to wide scale and sustainable implementation of new ideas.

S.665-668

In this panel, we will address questions on software engineering education that are voted on by the conference attendees in advance of the session. A sample set of questions will be posted. Conference attendees who will then vote on the questions to be addressed by the panel will augment these. After the panellists provide answers to a given question, the discussion will be opened up for follow-up questions or comments by the conference attendees.