**Session Date/Time:** 21 Mar 2024 07:15

```markdown
# TDD - BGP as the dumptruck

## Summary

This session delved into the multifaceted nature of BGP, exploring its evolution, extensibility, and the challenges of using it as a "dump truck" for various data types. The discussion covered topics ranging from BGP's core functionality and extensibility mechanisms to the security concerns and the potential for future improvements, including the possibility of BGP over QUIC.

## Key Discussion Points

*   **BGP's Extensibility:** BGP has remained relevant due to its extensibility, which allows for incremental deployment of new features without breaking existing implementations.
*   **The "Dump Truck" Analogy:** BGP is often used to carry a variety of data beyond its original purpose (IP reachability), leading to the "dump truck" analogy. This raises concerns about performance, scoping, and operational complexity.
*   **Scoping Challenges:** A major challenge is ensuring that data carried in BGP is properly scoped and doesn't leak into unintended contexts, particularly the internet.
*   **Error Handling:** BGP's original error handling mechanism, which involved resetting sessions upon encountering a malformed attribute, was overly aggressive. RFC 7606 introduced a more resilient approach based on the principle of minimum necessary force.
*   **TLVs and Extensibility:** BGP's extensibility relies heavily on TLVs (Type-Length-Value tuples). However, the lack of a uniform set of conventions for TLVs and the limited code point space pose challenges.
*   **Security Concerns:** BGP is vulnerable to various security threats, including prefix hijacking and AS path forgery. BGPsec aims to address these issues but faces deployment challenges due to its complexity and computational cost.
*   **BGPsec Deployment Challenges:** Concerns about PKI complexity, the expense of cryptographic operations, and the fact that BGPsec doesn't solve all security problems hinder its widespread adoption.
*   **Future Directions:** The session explored potential future directions for BGP, including:
    *   BGP over QUIC to improve transport efficiency and provide better framing.
    *   Building application profiles for resiliency.
    *   Improving filtering capabilities to allow operators to discard unwanted attributes.
    *   Schema-driven data structures (e.g., using CBOR) to improve data representation and validation.
    *   Better trust domain management within the protocol.

## Decisions and Action Items

*   **Elliot, Jeff and Sava to discuss a future conversation about Scion and BGP5.**
*   **Jeff, Sava, Ying Zheng, and Alvaroetana to engage in future discussion with Brian to see if there's future direction for the group's work in BGP over Quick.**

## Next Steps

*   The BGP over QUIC working group is pursuing the deployment.
*   The IETF IDR working group will work to enable more secure filtering mechanisms.