W3C Distributed Tracing Working Group Virtual Presence Workshop, March 2020

Part of Developers

Author(s) and publish date

By:
Published:

10 of the meeting participants displayed through the Zoom application

{Michele Mancioppi is missing on the screenshot)

This past week, the W3C Distributed Tracing Working Group workshop was held as a virtual presence event. As the Trace Context specification is already finalized, the hot topic of this workshop was the Correlation Context specification. We also started the next update of the Trace Context specification and making progress on it.

We had lighter than usual attendance - 11 people from Dynatrace, Google, Instana, Microsoft, LightStep, Uber Technologies, and W3C. “Virtual presence” nature of the event also changed the format - we had 3 hours meetings for 3 days. This format was well-received and we discussed how some learnings can be used in a group regular sync up happening every other week.

As the group charter reached the originally planned milestones, redefined milestones were proposed in the updated charter document.

We discussed again that correlation context specification needs to pick up speed. There are many requests to lock down the format and W3C recommendation would be the best place for it. Correlation context naming decision, leaning towards using Set-Cookie-like format.

As more systems implement the Trace Context specification, we had an update on what are the challenges of the implementation and sharing the same headers between different parties. In general, there were no red flags discovered. Some topics discussed were:

  • Use of tracestate. Dynatrace and Instana are actively using tracestate to communicate the relationships between requests seen by them while they may have transitioned through other systems.
  • As the group is very concerned about privacy and security, we have discussed that there were no new issues discovered so far compared to the old way of sending the same distributed context information using the vendor-specific headers.
  • Breaking other’s traces by clearing out tracestate or regenerating traceparent was discussed as a threat as everybody is using the same header. Interestingly, a real-life use case of traces cross-linking was hinted on during the discussion.

We will start to publish educational notes in our distributed tracing repository. We hope it will help improve implementers’ experience and we can publish docs more agile on faster cadence.

Please join our future meetings if you are interested in the topics we discuss. Or ask your questions and participate at any time.

All notes: W3C Distributed Tracing Working Group Remote Workshop (March 2020).

Session recordings for days 2 and 3.

Previous workshop report: Seattle, November 2019.

(post created by the Distributed Tracing Working Group)

Related RSS feed

Comments (0)

Comments for this post are closed.