anrw

Summary

The Applied Networking Research Workshop (ANRW) session covered a range of topics including IoT device location impact, programmable network infrastructure, and reflections on active network measurement in academia. Key themes included the challenges of DRAM scaling, the importance of representative network infrastructures for research, and ethical considerations in network measurement.

Key Discussion Points

DRAM Scaling Limitations: The keynote speaker discussed the flattening cost per bit of DRAM, limited latency improvements, and approaching signaling limits which will impact future computer architecture and application design. Three types of memory (latency-optimized, capacity-optimized, and bandwidth-optimized) are expected to emerge.
IoT Location Impact on MUD: A presentation explored how user-defined location impacts IoT device network behavior, specifically domain usage. It highlighted the need for MUD (Manufacturer Usage Description) to account for location, and a DNS ECS extension was proposed as a solution.
Programmable Infrastructure for Networking Research: This presentation showcased Peanut, a programmable infrastructure for network research at UC Santa Barbara, combining active and passive measurements on a campus network. It addressed the need for representative data and emphasized the importance of ethical data collection.
Reflections on Active Network Measurement: This talk discussed the challenges and ethical considerations involved in active network measurement. It proposed a community-governed infrastructure to alleviate the burden on individual researchers.

Decisions and Action Items

Call for Experiment Submissions (Peanut): The presenters of Peanut encouraged researchers to submit experiments for execution on their platform, providing email and website information for contact.
Community-Governed Measurement Infrastructure (Tobias): The presenter is building a community-governed infrastructure and will reach out to appropriate organizations to take ownership.

Next Steps

The audience are encouraged to read the papers presented at the workshop, accessible on the ANRW23 website.
The audience can follow up with the presenters with further questions.
Explore the Peanut platform and the speaker, Roman, is contactable by email or through his website.

Session Date/Time: 24 Jul 2023 20:00

anrw

Summary

This meeting focused on the future of the Internet in the next 20 years, with presentations and discussions covering network performance, refugee crisis mapping using Internet measurements, and research opportunities in security and privacy. The panel discussion explored desired attributes of the future Internet, including ubiquity, sustainability, decentralization, and security, and addressed potential challenges from political opposition and societal collapse.

Key Discussion Points

Network Performance: Simon Mombauer presented research on the impact of TCP options (window scaling, selective acknowledgments, ECN), QUIC implementations (quiche, aioquic), and CDNs on throughput. Window scaling proved crucial, while QUIC implementations showed significant performance differences. Further measurements with longer durations are planned to understand these variances.
Refugee Crisis Mapping: Tal Rahan discussed using Internet measurements (Google Maps traffic, mobile device usage) to map the Ukrainian refugee crisis. Mobile device usage in Poland correlated strongly with refugee influx. The methodology aims to complement UN data using publicly available data for privacy preservation.
Security and Privacy Research Opportunities: Chris Wood presented research opportunities in the security and privacy space, focusing on the need for formally verified implementations and improved specifications.
Multiparty Computation (MPC): Identified as a hot research area, particularly around private heavy hitters and composition with differential privacy.
Anonymous Credentials: Discussed as an ongoing research area with challenges in post-quantum security and practical deployment.
Decentralization: Panel discussion explored the meaning of decentralization in the context of Internet control and the increasing influence of large corporations. It raised concerns about data consolidation.
End-to-End Encryption: The panel debated the feasibility and challenges of achieving true end-to-end encryption in a decentralized and mobile environment, including identity management and potential government opposition.
Impact of Quantum Computing: The impact of quantum computing on Internet security and the progress made in deploying post-quantum cryptography was discussed.
Fractured Internet: A pessimistic view was raised that the Internet may fragment into distinct regions with differing access to information.
Security Definition: It was argued there is a lack of shared understanding in the security problem space. A technical solution might not solve societal problems and laws might be necessary.

Decisions and Action Items

Further Measurements: The presenter of the TCP/QUIC performance talk will conduct further measurements with higher iteration counts.

Next Steps

Discussions and explorations of these topics may continue in related IETF working groups or research communities.

Session Date/Time: 24 Jul 2023 22:30

# anrw

## Summary

The anrw session featured presentations on various network-related topics, including lowering barriers to working with public reliable data, DNS integrations, repeatable name resolution, multi-homed Hypergiant collaboration, and anomaly detection in large-scale BGP VPN networks. Each presentation was followed by a Q&A session.

## Key Discussion Points

*   **Lowering Barriers to Public Reliable Data:**  Discussion centered around inconsistencies in publically available data from regional internet registries (RIRs) such as the use of different attributes for similar information and lack of historical data.  The presentation introduced a consolidated data format with identifiers and the complement of missing data.  Clarification was provided that the presented data came from public IRR data, not live whois queries.
*   **DNS Integrations:** Explored different types of DNS integrations (DNS-based and server-based) and challenges related to synchronization, domain lifecycle, and control. A call for collaboration was made to establish best practices for different types of DNS integrations. The suggestion of a BoF was made.
*   **Repeatable Name Resolution:**  Presented a method for discovering all DNS dependencies and identifying inconsistencies by building a complete dependency tree through building their own resolver. Zone cut discovery was discussed as well as querying consistent sets of TLD name servers to avoid overburdening them. Use cases and availability of data sets were covered.
*   **Multi-Homed Hypergiant Collaboration:** Discussed the problem of suboptimal server selection by Hypergiants for non-directly connected ISPs and proposed a collaboration model where ISPs provide additional information to Hypergiants. Retrospective simulation using real ISP data demonstrated potential improvements. Discussion focused on how selection could change the route.
*   **Anomaly Detection in BGP VPN Networks:** Showcased a framework (Daisy) for detecting anomalies in large-scale BGP MPLS and BGP SRv6 VPN networks, focusing on customer impact. Different data sources (IPFIX, BMP, YANG Push) were used, and customer behavior was profiled. The use of machine learning was discussed to build new checks.

## Decisions and Action Items

*   **DNS Integrations:**  Explore the possibility of a BoF to discuss best practices for DNS integrations with different communities and organizations.
*   **Repeatable Name Resolution:** The presenter will make the data set available for further analysis and has received feedback to create a tool suitable for domain admins.

## Next Steps

*   Further exploration of collaborative opportunities for establishing DNS integration best practices.
*   Analyzing gathered data with repeatable name resolution and identify new interesting questions.
*   Continue working on the anomaly detection framework, focusing on performance and the impact of different network events.