Debugging QUIC and HTTP/3 with qlog and qvis

Marx, Robin; Piraux, Maxime; Quax, Peter; Lamotte, Wim

doi:10.1145/3404868.3406663

Cited by 21 publications

(8 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We shape delay using tc netem on the interconnecting links of the veth pairs on each end-host and configure a steady bandwidth of 10 Mbps 2 on the client's ingress interface. For measuring RTTs, we create QUIC traffic using aioquic [21] at the QUIC hosts (blue) and capture the output of our tracker, the qlog [23] information of aioquic, and the traffic at the client's ingress interface, which we subsequently analyze using Spindump [10]. While the Spindump results provide a groundtruth regarding the accuracy of plain spin bit measurements, the qlog data serves as a groundtruth for the actual RTT.…”

Section: Discussionmentioning

confidence: 99%

Tracking the QUIC spin bit on Tofino

Kunze

Sander

Wehrle

et al. 2021

Proceedings of the 2021 Workshop on Evolution, Performance and Interoperability of QUIC

View full text Add to dashboard Cite

QUIC offers security and privacy for modern web traffic by closely integrating encryption into its transport functionality. In this process, it hides transport layer information often used for network monitoring, thus obsoleting traditional measurement concepts. To still enable passive RTT estimations, QUIC introduces a dedicated measurement bit -the spin bit. While simple in its design, tracking the spin bit at line-rate can become challenging for software-based solutions. Dedicated hardware trackers are also unsuitable as the spin bit is not invariant and can change in the future.Thus, this paper investigates whether P4-programmable hardware, such as the Intel Tofino, can effectively track the spin bit at line-rate. We find that the core functionality of the spin bit can be realized easily, and our prototype has an accuracy close to softwarebased trackers. Our prototype further protects against faulty measurements caused by reordering and prepares the data according to the needs of network operators, e.g., by classifying samples into pre-defined RTT classes. Still, distinct concepts in QUIC, such as its connection ID, are challenging with current hardware capabilities.

show abstract

Section: Discussionmentioning

confidence: 99%

Tracking the QUIC spin bit on Tofino

Kunze

Sander

Wehrle

et al. 2021

Proceedings of the 2021 Workshop on Evolution, Performance and Interoperability of QUIC

View full text Add to dashboard Cite

show abstract

“…We define the transfer time as the delay between the first QUIC packet sent by the client and the last QUIC packet received by the client containing the CONNECTION_CLOSE frame. In addition, we generate QLOG files [25] at client and server sides to get a full view of the connection and the internal state of the implementations. We first experiment with 2-path network scenarios where both paths share the same characteristics.…”

Section: Discussionmentioning

confidence: 99%

The Packet Number Space Debate in Multipath QUIC

Coninck¹

2021

Preprint

View full text Add to dashboard Cite

With a standardization process that attracted many interest, QUIC can been seen as the next general-purpose transport protocol. Still, it does not provide true multipath support yet, missing some use cases that MPTCP can address. To fill that gap, the IETF recently adopted a multipath proposal merging all the proposed designs. While it focuses on its core components, there still remains one major design issue in the proposal: the number of packet number spaces that should be used. This paper provides experimental results with two different Multipath QUIC implementations based on NS3 simulations to understand the impact of using one packet number space per path or a single packet number space for the whole connection. Our results suggest that using one packet number space per path makes the Multipath QUIC connection more resilient to the receiver's acknowledgment strategy.

show abstract

“…On the other hand, DoQ (solid cyan line) differs drastically from the expected handshake of 1 RTT : With around 20% of measurements showing an RTT of 1, DoQ converges to 2 RTTs at the 60 th percentile; hence, roughly 40% of DoQ measurements require more than 2 RTTs, which is twice as much as expected in comparison. To investigate this, we analyze the qlog [49] outputs recorded during our response time measurements, which enable us to analyze the packet exchanges in detail. Using the qlogs, we confirm that the response time measurements are not affected by QUIC's Version Negotiation feature, as we use the previously negotiated QUIC Version of the cache warming session for the handshake of the subsequent DNS response time measurement session (see § 2).…”

Section: Response Timesmentioning

confidence: 99%

One to Rule them All? A First Look at DNS over QUIC

Kosek,

Doan,

Granderath

et al. 2022

Preprint

View full text Add to dashboard Cite

The DNS is one of the most crucial parts of the Internet. Since the original DNS specifications defined UDP and TCP as the underlying transport protocols, DNS queries are inherently unencrypted, making them vulnerable to eavesdropping and on-path manipulations. Consequently, concerns about DNS privacy have gained attention in recent years, which resulted in the introduction of the encrypted protocols DNS over TLS (DoT) and DNS over HTTPS (DoH). Although these protocols address the key issues of adding privacy to the DNS, they are inherently restrained by their underlying transport protocols, which are at strife with, e.g., IP fragmentation or multi-RTT handshakes -challenges which are addressed by QUIC. As such, the recent addition of DNS over QUIC (DoQ) promises to improve upon the established DNS protocols. However, no studies focusing on DoQ, its adoption, or its response times exist to this date -a gap we close with our study. Our active measurements show a slowly but steadily increasing adoption of DoQ and reveal a high week-over-week fluctuation, which reflects the ongoing development process: As DoQ is still in standardization, implementations and services undergo rapid changes. Analyzing the response times of DoQ, we find that roughly 40% of measurements show considerably higher handshake times than expected, which traces back to the enforcement of the traffic amplification limit despite successful validation of the client's address. However, DoQ already outperforms DoT as well as DoH, which makes it the best choice for encrypted DNS to date.

show abstract

Debugging QUIC and HTTP/3 with qlog and qvis

Cited by 21 publications

References 28 publications

Tracking the QUIC spin bit on Tofino

Tracking the QUIC spin bit on Tofino

The Packet Number Space Debate in Multipath QUIC

One to Rule them All? A First Look at DNS over QUIC

Contact Info

Product

Resources

About