Session Date/Time: 01 Jun 2022 15:00

PANRG

Summary

This meeting featured a presentation on SCION, a path-aware internet architecture, and an extensive discussion on its potential role within the IETF/IRTF. The core of the discussion revolved around how SCION, currently presented as a holistic system, could be analyzed in terms of its individual components, their maturity, and their relationship to existing internet protocols. A key outcome was the group's agreement that PANRG is an appropriate venue to delve into a "SCION component analysis" to guide future work.

Key Discussion Points

SCION Overview: Corinne Carter (SCION Association) presented an overview of SCION, emphasizing its design for scalability, control, and isolation on next-generation networks. Key features include inter-domain path awareness, trust heterogeneity through Isolation Domains (ISDs), separate control and data planes with their own PKI, and stateless routers.
Current Deployments: SCION is in production with 7 ISPs and 5 trial deployments across various sectors (finance, education, government, healthcare), demonstrating practical applicability. SCIONLab provides an R&D network for testing.
Motivation for Standardization: The SCION team highlighted growing deployments and demand from early adopters for standardized implementations, which would facilitate acceptance and ensure interoperability. They believe SCION addresses questions raised in RFC 9217 ("Current Open Questions in Path-Aware Networking") and meets criteria for successful protocols (RFC 5218).
Feedback on Informational Draft: While the draft (draft-copter-panrg-scion-overview) was acknowledged as well-written and a useful high-level introduction, a sense of those present indicated that for IETF work, more emphasis on protocol bits and building blocks is needed, rather than solely an architectural overview.
Reusing vs. Inventing Protocols: A significant point of discussion, led by Myria Peters and Spencer Dawkins, questioned why SCION introduces new protocols (e.g., for PKI, data plane) instead of adapting or extending existing IETF standards (e.g., current PKI, segment routing).
- SCION proponents (Adrian Perrig, Nicola Rustignoli) cited the need for additional security mechanisms (cryptographic identifiers) not present in segment routing, and the fundamental difference in SCION's trust model for inter-domain communication, which assumes non-trusting entities between ISDs, unlike intra-domain environments. Header efficiency was also mentioned.
Obstacles to Deployment (RFC 9049): Spencer Dawkins asked the SCION team to consider RFC 9049, which lists common pitfalls for path-aware networking protocols. The SCION team confirmed awareness of this RFC and stated their belief that SCION's design avoids these issues.
Application Adoption: Suresh Krishnan inquired about how applications would use SCION (native vs. gateways). Adrian Perrig noted current reliance on gateways, but ongoing research into native application support via libraries, stack bundling, and auto-discovery, with concepts like extending "Happy Eyeballs" and insights flowing into TAPS.
Transition Mechanisms: Nicola Rustignoli asked if detailing experimental transition mechanisms (e.g., extending Happy Eyeballs, auto-discovery) would be useful for the draft. Brian Trammell, as chair, suggested these topics are "bread and butter" for PANRG and would be excellent for dedicated research group documents.
Deployment Experience and Negative Results: Myria Peters and Brian Trammell emphasized the value of documenting SCION's real-world deployment experiences, including challenges faced and what didn't work (negative results), as this is a unique contribution that other path-aware proposals lack.
System vs. Components: Colin Perkins (IRTF Chair) reiterated that the IETF excels at standardizing components rather than entire systems. He stressed the need to identify SCION's distinct components, their individual maturity levels, and how they might fit into existing IETF or IRTF work streams.
Interoperability and Versioning: Suresh Krishnan raised important practical considerations for future standardization, including how to ensure interoperable gateway implementations beyond a single reference, and robust versioning mechanisms for the protocol.

Decisions and Action Items

Decision (PANRG): A poll of those present indicated strong interest in a "SCION component analysis" within PANRG. The group believes this would be a useful product of the research group.
Action Item (SCION Presenters): The SCION team is requested to come prepared for IETF 114 to initiate a discussion (via presentation or an initial draft) on the "component analysis" of SCION. This analysis should aim to:
- Identify and decompose SCION into its distinct components.
- Detail the similarities and differences between SCION components and existing internet architecture elements.
- Explore the implications and feasibility of deploying only specific SCION components.
- Analyze how existing IP stack protocols or components could be adapted or evolved to replace missing parts of the SCION system if only a subset were deployed.
- Document deployment experiences, including challenges and what did not work, to inform broader path-aware networking research.
Action Item (PANRG Co-Chairs): Schedule significant time on the PANRG agenda at IETF 114 to continue this discussion on SCION's component analysis. The chairs will also be prepared to schedule additional interim meetings between IETFs to foster ongoing collaboration and discussion on these topics.

Next Steps

Continued Discussion at IETF 114: The primary next step is to continue the in-depth discussion on SCION's component analysis at the PANRG session during IETF 114 (Philadelphia, July). SCION proponents should come prepared to lead this discussion, focusing on the questions outlined in the action item.
Mailing List Engagement: SCION proponents are encouraged to use the PANRG mailing list for ongoing questions, sharing early thoughts, or seeking feedback on their analysis.
Future Documentation: Depending on the progress of the component analysis, future informational drafts could emerge focusing on:
- Detailed technical specifications of individual SCION components identified as mature enough for IETF consideration.
- Analysis of SCION's transition mechanisms and incremental deployment strategies.
- Implementation reports documenting real-world deployment challenges and lessons learned.
- A comprehensive "gap/reuse analysis" that specifically details how existing protocols could be adapted or extended to incorporate SCION's principles.
IRTF/IETF Pathfinding: The goal of these discussions is to identify which parts of SCION are mature enough for immediate IETF standardization (engineering) and which require further research (IRTF), ultimately aiming for concrete proposals for appropriate IETF working groups in the future (potentially targeting the March 2023 meeting).
Current Overview Draft: The existing informational draft (draft-copter-panrg-scion-overview) remains a valuable background document for participants in this ongoing discussion.