**Session Date/Time:** 18 Mar 2024 03:00

# maprg

## Summary

This IRTF meeting featured several presentations on network research topics. Key discussions included a BPF implementation of IPv6 extension headers, updates on spin bit measurements in QUIC, improving outage detection using multiple probing protocols, pruning IRR entries with RPKI, anycast polarization in the wild, and mobile app filtering in India. The presentations were followed by questions and discussions among the attendees.

## Key Discussion Points

*   **BPF Implementation of IPv6 Extension Headers (PDium):**
    *   Exploration of user EVPF in networking.
    *   Implementation of IPv6 extension headers in EVPF.
    *   Performance analysis of the EBPF implementation against a traditional kernel implementation.
    *   Optimization potential of the EBPF program.
*   **Spin Bit Measurements in QUIC (Ike):**
    *   Update on previous work regarding the spin bit in QUIC for round trip time measurement.
    *   Assessment of spin bit usage and accuracy in web measurements.
    *   Investigation into disabling compliance and RTT overestimation.
    *   Impact of different stacks and configurations on spin bit performance.
    *   Discussion on the use of new connection ID and retire connection ID frames, as well as ping frames, and their effect on spin bit measurement accuracy.
*   **Improving Outage Detection with Multiple Probing Protocols (Manasini):**
    *   Limitations of ICMP probing for outage detection.
    *   Potential improvements by adding TCP and UDP probing.
    *   Analysis of historical block availability using ICMP data and TCP/UDP scan data.
    *   Evaluation of host and block availability and reliability.
    *   Discussion of the value of slash 24 metrics for outage detection.
    *   Questions regarding the lack of ipv6 in the research.
*   **Pruning IRR Entries with RPKI (TJ):**
    *   Using RPKI to improve the quality of IRR entries by removing stale entries.
    *   Characterizing inconsistent IRR objects and applying machine learning techniques.
    *   Comparison with other techniques, such as IRRd4.
    *   Challenges in reducing false negatives.
*   **Anycast Polarization in the Wild (Rishvi):**
    *   Identification of key reasons behind anycast polarization, including multiple backbone problems and regional routing issues.
    *   Investigation of polarization problems across known prefixes.
    *   Working with a CDN provider to mitigate the impact of polarization using routing changes.
    *   Discussion of transit connectivity and peering problems.
    *   Discussion on automated detections and manual optimization for routing changes.
*   **Mobile App Filtering in India (Apoorv):**
    *   Study of mobile app filtering practices in India following the banning of Chinese apps.
    *   Investigation into geoblocking techniques used by app publishers.
    *   Analysis of source IP and CDN edge server filtering.
    *   Reverse engineering of apps to identify the use of Android telephony manager package to access SIM card information.
    *   Discussion of complying with the order.

## Decisions and Action Items

*   **Spin Bit Measurements in QUIC:** Vic will follow up with Ike to discuss questions about the spin bit.
*   **Improving Outage Detection with Multiple Probing Protocols:** Presenter will return with results of future work.

## Next Steps

*   **BPF Implementation of IPv6 Extension Headers (PDium):** Continue optimization of the EBPF program and analysis of performance overhead.
*   **Spin Bit Measurements in QUIC:** Continue investigation into mandatory disabling, RTT overestimation, and stack-specific behavior in a test bed study.
*   **Pruning IRR Entries with RPKI:** Continue working on reducing false negatives in the machine learning model.
*   **Anycast Polarization in the Wild:** Further research into automated detection and optimization strategies for routing changes.
*   **Mobile App Filtering in India:** Continue research into app filtering to identify the potential reasons.