Markdown Version | Session Recording
Session Date/Time: 24 Jul 2024 16:30
maprg
Summary
This meeting of the Measurement Analysis for Protocols Research Group (maprg) featured presentations on a variety of topics related to internet protocol measurement and analysis. Discussions covered HTTP/3 prioritization, BGP looking glass data, network link conditioning for protocol testing, YouTube streaming over geo-satellite networks, NAT64 deployment, IPv6 honeypots, and post-quantum DNSSEC.
Key Discussion Points
- HTTP/3 Prioritization: Joris Herbots presented findings on the heterogeneity and unpredictable behavior of Extensible Prioritization Scheme (EPS) implementations in HTTP/3, highlighting the need for full support and better developer APIs.
- BGP Looking Glass Data: Pascal Hen discussed a project to collect and analyze BGP configuration data from looking glasses, providing insights into AS diversity and routing policies.
- Network Link Conditioning: Tommy Pauly emphasized the importance of protocol testing under simulated network conditions, showcasing how network link conditioning can reveal performance issues in deployments like HTTP/3-based forward proxies.
- YouTube over Geo-Satellite Networks: Jamal Leo presented an analysis of YouTube quality of experience (QoE) over geo-satellite networks, finding that traffic shaping can be tricky and that the achieved throughput is lower than expected, partially due to the long propagation delay.
- NAT64 Deployment: Amanda presented results indicating limited deployment of public DNS64 resolvers and NAT64 clients (based on RIPE Atlas probes), and showed that most NAT64 gateways are not publicly accessible. Jen Linkova and Stuart Cheshire provided important context around the use of DNS64, recommending caution due to the importance of coupling between DNS64 resolvers and the network, and suggesting that NAT64 is important and distinct from DNS64.
- IPv6 Honeypots: Phil Roberts (presentation completed by Leslie Daigle after Phil's connection dropped) discussed the challenges of deploying IPv6 honeypots, particularly in attracting attention due to the vastness of the IPv6 address space. Wes Hardaker suggested GCA share data on IPv4 attacks to provide a comparison benchmark.
- Post-Quantum DNSSEC: Jason Gertsen presented results from experiments with post-quantum cryptographic (PQC) algorithms in DNSSEC, highlighting the trade-offs between security and performance due to the large signature sizes of PQC algorithms.
Decisions and Action Items
- Action Item: Phil Roberts to contact Wes Hardaker re: sharing IPv4 Honeypot data for benchmark comparisons.
Next Steps
- Continue research and analysis in the areas of protocol measurement and performance.
- Explore collaborative efforts to share network settings and profiles for simulated testing.
- Investigate the use of Merkel trees for compressing PQC signatures in DNSSEC.