IRTF maprg agenda for IETF-114 (Philadelphia)  
Date: Friday, July 29, 12:30-14:30 Session II UTC-4 Webex link:
https://meetings.conf.meetecho.com/ietf114/?group=&short=&item=1 Room:
Liberty C

Overview & Status - Mirja (onsite) & Dave (remote) (5 min)  
IRTF Note-well: https://irtf.org/policies/irtf-note-well-2019-11.pdf

Abstracts:
https://datatracker.ietf.org/meeting/114/materials/agenda-114-maprg-05

Materials: https://datatracker.ietf.org/meeting/114/session/maprg

Heads-up talk: Internet Performance in the 2022 Conflict in Ukraine: An
Asymmetric Analysis - Tal Mizrahi (10 mins)  
https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-internet-performance-in-the-2022-conflict-in-ukraine-00.pdf

*   Max Franke: starlink info?
    *   Tal: probably improved connectivity, we don't have stats

*   Max: stats on search trends?
    *   Tal: we focused on search rate, not the content

*   Ian Williams: would you have disabled the in-country-only redirect
    in google's shoes to avoid leaking this data?
    *   Google not the only one, we don't think it affected results much

*   Tianji: any info on the stats for damaged infrastructure traffic
    offloading to starlink? Aggregate stats on that?
    *   Related to the first question, we don't have stats for starlink.

A Look at QUIC Use - Geoff Huston (15 mins)  
https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-a-quick-look-at-quic-geoff-huston-00.pdf

*   Shivan Sahib: (brave) -- account for ad-blocking? that's 35% of the
    web as of a few years ago
    *   this is world seen by ads, not the world. i get very little data
        from anyone blocking ads

*   Ben Schwartz: are you killing quic from server side before the 2nd
    request?
    *   that's an nginx question, we're doing whatever it does
    *   also some impact from delay variation on the browser side. we're
        gonna change the timer to do a sequence of delayed fetches
    *   ben: try http flows or close the h2 connection as soon as you've
        sent the response, the problem you're seeing is reuse of
        connection probably since a socket is already open with h2

*   Ben again: for https IP hints, only necessary if your target name is
    not '.', the magic default value meaning not doing anything funky
    with redirect

Configanator: A Data-driven Approach to Tackle Network Diversity with
Heterogeneous Configurations - Naseer, Usama (15 mins)

*   https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-configanator-a-data-driven-approach-to-tackle-network-diversity-with-heterogeneous-configurations-00.pdf
*   https://www.usenix.org/system/files/nsdi22spring\_prepub\_naseer.pdf
*   https://www.usenix.org/conference/nsdi22/presentation/naseer

*   Tianji: Improvement is negative in some cases, up to 20% of the
    time. Why?
    *   this is an online search algorithm, during the search some
        experimental points will test adverse impacts as well. But the
        algorithm will converge to net better. The paper goes into some
        more details

*   Ben Schwartz: have you looked at game theoretic implications here?
    How much gain is someone else's loss?
    *   good question, the paper goes into some fairness examination but
        we haven't looked at game theoretic considerations yet, if we
        think of a future where everyone runs these

Active TLS Stack Fingerprinting: Characterizing TLS Server Deployments
at Scale - Markus Sosnowski (15 mins)  
https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-active-tls-fingerprinting-00.pdf

*   Dave Plonka: how do you determine ground truth about which cdn it
    was when you compare the identifiability of the cdn fingerprints?
    And did you look at multi-cdn?
    *   didn't look at multi-cdn
    *   ground truth we verified by e.g. asking for a cloudflare cert,
        if they were able we assumed they were actually cloudflare.

Analyzing the Influence of Resource Prioritization on HTTP/3 HOL
Blocking and Performance - Constantin Sander (remote) (15 mins)  
https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-analyzing-the-influence-of-resource-prioritization-on-http3-hol-blocking-and-performance-00.pdf

Measuring the Accessibility of Domain Name Encryption and its Impact on
Internet Filtering - Nguyen Phong Hoang (remote) (15 mins)

*   https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-measuring-the-accessibility-of-domain-name-encryption-and-its-impact-on-internet-filtering-00.pdf
*   paper: https://arxiv.org/pdf/1911.00563.pdf
*   data:
    https://homepage.np-tokumei.net/publication/publication\_2022\_pam/
*   ref'd in presentation: https://gfwatch.org/
*   Dave Plonka: do you know of work that classifies the type of
    censorship? e.g. malware or explicit content vs. political?
    *   most censorship we found was from political content.
    *   work from great firewall study lists what gets blocked in china

*   Patrick McManus: nuance comment: censorship case where blocking is
    applied still preserves confidentiality, right? This is strictly
    better than the prior situation of plaintext analysis
*   Patrick 2: support the call for universal ECH, good take. Wheels
    turn slowly but can get there.
    *   yes, this can cause some new problems (e.g. malware scanning
        getting harder) but in the long run it'll be a net win,
        hopefully universal deployment will help address problems of
        censorship and monitoring.

Measuring the Availability and Response Times of Public Encrypted DNS
Resolvers - Nick Feamster (remote) (15 mins)  
https://datatracker.ietf.org/meeting/114/materials/slides-114-maprg-measuring-the-availability-and-response-times-of-public-encrypted-dns-resolvers-00

*   measurement tool: https://noise-lab.github.io/dns-measurement/