Title: An Analysis of QUIC Connection Migration in the Wild
Authors: Buchet Aurélien, Pelsser Cristel (UCLouvain)
Scribe: Yuntao Zhao (Xiamen University)
Introduction
The QUIC protocol has gained significant traction in recent years as a fast, secure transport layer for web traffic. One of its novel features is connection migration, which allows an ongoing QUIC connection to seamlessly switch a host’s IP address without restarting the session. This capability can enable fast handoffs in mobile scenarios (similar in spirit to Multipath TCP) and can serve privacy/security purposes by thwarting on-path tracking via IP changes. Despite these potential benefits, it has been unclear to what extent web servers in the wild actually support connection migration. To address this gap, the authors conduct an Internet-wide empirical study to evaluate the current deployment of QUIC’s connection migration feature.
Key Idea and Contribution
- Internet-Wide Scanning: The authors performed comprehensive scans of the IPv4 address space (and extensive IPv6 hitlists) to identify QUIC-capable servers, reproducing state-of-the-art methods (e.g., using ZMap) from prior QUIC studies. This provided a broad set of target servers that speak QUIC, forming the basis for evaluating migration support.
- Migration Testing Tool: They developed a new client-side tool to actively test QUIC connection migration on those servers. The tool initiates a QUIC connection and then attempts to migrate it (for example, by changing the client’s network interface or port) to see if the server continues the session. This innovation enabled the first large-scale probing of servers’ support for connection migration.
- Quantifying Deployment Support: Using the above approach, the paper quantifies what fraction of QUIC-supporting servers also support connection migration, and it compares these findings against earlier studies of general QUIC adoption. The work provides the first quantitative insight into how widely QUIC’s migration feature has been adopted in practice, complementing existing literature on QUIC deployment.
Evaluation
The results present a clear snapshot of QUIC connection migration support (or lack thereof) across the Internet. On one hand, QUIC itself is rapidly expanding – the scans found nearly five times more responsive QUIC servers than reported in 2022 studies, indicating a dramatic growth in QUIC adoption. On the other hand, support for the connection migration feature is far from universal. Many high-profile web destinations and major providers have not yet enabled connection migration in their QUIC deployments. The measurements showed that only a subset of servers successfully handled the client’s address change mid-connection. Moreover, the support is highly skewed towards certain providers: for instance, in the IPv6 scans, out of about 693k targets that supported migration, over 654k belonged to a single hosting provider (Hostinger), while most others offered little to no support. This suggests that as of now, QUIC’s migration capability is confined to specific corners of the Internet infrastructure. The authors note that despite the promising use cases for performance and privacy, the deployment of connection migration among “big” QUIC players is not yet widespread. They advocate for continued longitudinal scans to track adoption over time and emphasize the need to investigate why many servers opt not to support migration. Understanding these barriers (be they technical challenges, security concerns, or lack of demand) could help in promoting broader enablement of the feature. It’s also worth noting that the study’s approach focused on client-side initiated migration (using the same interface/port); future work could extend to scenarios like server-preferred address migration to fully examine the feature’s deployment.
Personal Thoughts
This measurement study provides a timely reality check on the state of QUIC’s advanced features. By combining large-scale scanning with a novel migration-testing tool, the authors shine light on the fact that QUIC’s connection migration — while theoretically beneficial — has seen limited real-world uptake so far. These findings are valuable to protocol engineers and network operators: even as QUIC usage grows, certain enhancements like migration might lag in adoption due to practical considerations. The study, by filling a gap in the literature, offers a baseline for the community to understand current deployment and to monitor progress going forward. As the authors suggest, investigating the reasons behind providers’ reluctance to enable migration will be an important next step. It could be that concerns about complexity, stability, or security are impeding deployment; addressing these will be key to unlocking the full potential of QUIC’s mobility and privacy features. Overall, this work delivers a much-needed empirical perspective. In my view, it will spur both researchers and practitioners to pay closer attention to QUIC’s advanced capabilities and perhaps accelerate efforts to make connection migration more widely available.