The Security WG Call discussed the Q1 Security Workshop Agenda, the Threat Model document finalization, and the CDN Log Schema document. They discussed the need for a common understanding of log data across multiple CDN providers, emphasizing the importance of normalizing data fields for consistent analysis. The group focused on securing sensitive data in CDN logs. Brian Paxton emphasized the need for secure tokens or cookies to store session IDs, suggesting removal of redundant fields like session ID from the document. Glenn Deen agreed, advocating for using existing industry tokens. The group discussed the distinction between URL path and URL tokens, with Brian clarifying that parameters should not be included in the URL path. They also debated the relevance of application delivery network (ADN) terminology, ultimately deciding to remove it. The meeting concluded with plans to review and discuss further changes in the next call.
- Perform edits to the CDN logging schema: remove the generic UUID row, add/clarify request ID and session ID entries, and update tracked changes accordingly to reflect the agreed normalization approach.
- Follow up with Netflix/Bastien to confirm slide submission deadline and finalize their talk participation for the workshop.
- Perform follow-ups on outstanding workshop speaker confirmations and agenda details (complete remaining outreach and updates to the agenda spreadsheet).
- Present an overview and discussion on CDN leeching (beIN / live sports) at the workshop — prepare a 20-minute talk and be ready to field questions.
- Antitrust Statement & Security Workshop Agenda Finalization
- Kei outlines the agenda for the call, including discussing the agenda for the security workshop, the finalization of the Threat Model paper, and the discussing the CDN Log Schema document.
- The group then discusses the Security Workshop agenda at the Q1 meeting.
- Glenn Deen confirms his talk on SVTA Content Security work and the security needs around the SVTA edge.
- Kei and Glenn discuss the need for time adjustments to accommodate all presentations and discussions.
Threat Model Document Finalization
- Brian Paxton and Kei review the Threat Model document, making final changes and addressing outstanding issues.
- The group agrees to finalize the Threat Model document, considering the feedback and suggestions from the discussion.
- Brian Paxton plans to tidy up the document and address any remaining issues in future updates.
CDN Log Schema Document Discussion
- Brian Paxton presents changes to the CDN Log Schema document, including the addition of new fields and the clarification of objectives and scope.
- Mark Nakano and Brian Paxton discuss the importance of normalizing data across different CDN providers.
- The group debates the use of generic terms like UU ID and the need for more specific field names.
- Glenn Deen suggests using a key value or key ID to facilitate cross-log correlation and analysis.
- Brian Paxton and the group discuss the challenges of defining common log fields across different CDN providers.
- Mark Nakano emphasizes the need for consistent data governance to ensure accurate analysis and comparison.
Discussion on Token Security and Identifiers
- Brian Paxton emphasizes the need for secure tokens, suggesting they be part of a signed URL, cookie, or authorization header.
- Glenn Deen agrees but suggests focusing on using existing industry tokens rather than inventing new ones.
- Kei mentions the use of trace ID or request ID in systems like Akamai and Fastly, questioning if it can be used for the same purpose.
- Brian Paxton clarifies the complexity of the document, discussing the uniqueness of request IDs and their role in tracking requests.
Clarification on URL Path and Parameters
- Brian Paxton and Glenn Deen discuss the definition of URL path, confirming it does not include parameters after the question mark.
- Brian Paxton explains the request URL should contain the entire URL, including the domain, path, and any parameters.
- Glenn Deen agrees, noting that parameters are not part of the URL path but could be part of URL tokens.
- Brian Paxton suggests removing session ID from the document, as it should be securely within a token or cookie.
Session ID and Request ID Discussion
- Brian Paxton proposes adding a unique identifier for the playback session under the request ID, as session ID is not always passed.
- EZDRM mentions that Apple does not pass the CMC D session ID, complicating its inclusion.
- Brian Paxton suggests adding a comment that session ID is typically found within a secure token or cookie.
- Brian Paxton proposes tracking changes to session ID and request ID for future discussion.
Status Message and Application Delivery Network
- Brian Paxton introduces the status message field, indicating the HTTP status message from various servers.
- Glenn Deen is unfamiliar with the term “ADN” and suggests it might be specific to Sean McCarthy’s document on schema.
- Kei clarifies that ADN is an application delivery network, different from a CDN, which focuses on static content delivery.
- Brian Paxton proposes removing ADN from the document, as it is specific to CDN logs.
Final Remarks and Next Steps
- Brian Paxton suggests reviewing the document’s end, which includes sections on WAFs, ISP names, and client cities, to determine their relevance.
- Kei agrees and mentions the upcoming call before the meeting to discuss any last-minute changes.
- Brian Paxton thanks everyone for their participation and contributions.
- The meeting concludes.
Next Meeting: Tuesday, February 24, 2026 8am PST