perfSONAR User Q&A and Other Discussion

[Luke Fowler <>]

Greetings perfSONAR Users,

Thanks for all of you that participated in our recent user survey!  We received 82 responses from the perfSONAR community, including users at Universities, National Labs, Regional Networks, Government, Industry, and others.  We received responses from every continent!  Attached to this e-mail is a PDF containing the summary results from the survey.  We’ve also tried to capture and summarize the most popular/significant responses to the free-form sections of the survey below.  

This community feedback is extremely valuable to us!  We’ll be using your feedback as input to our project planning process at our upcoming developer face to face meeting next week and will be sharing more details on our future roadmap (based on the input we received in this survey!) after this meeting.

Thanks again for your interest and participation in the survey.  We of course welcome any additional feedback you have at any time!

(Apologies - some of the longer individual response selections in the attached summary PDF are truncated, please refer to the survey form for the full version of each response item:  https://docs.google.com/forms/d/1aQLiufmWLKvfjWyekKz5p_XsERdAU6ZzUMWoZFelZjc/ )

—Luke, on behalf of the perfSONAR team

Summary of free-form questions: 

New Features:

- easier management of meshes is important

- multiple mentions of running perfSONAR directly on router/switches

- Data visualization is an area people think there is room for improvement

- more analysis / guided troubleshooting / troubleshooting tips are desired

- integration with 3rd party data analysis tools (e.g. Grafana) was mentioned by multiple people

- thresholds for results was mentioned (presumably for alarming purposes)

-  auto-discovery of test configuration

- cloud deployment was mentioned a couple of times

Other Enhancements:

- Many of the above were repeated in this section

- Fewer ports used (firewall configuration ease)

- Support for TWAMP and other OA&M protocols common on CPE devices

- Better support for “non-expert” / “non-network-engineer” users

- Integration with external monitoring systems

- Better support for NAT (cloud use case)

- More precise timing for short links would be good to investigate (latency?)

- regression testing before releases are important

Gaps:

- Many repeats of the above items

- Documentation improvement is important

- Demo use cases / POC examples

- video training is desired

- Configuration / maintenance automation could be improved

- more web-based setup (less linux knowledge needed)

- Better debugging when tests are not working is desired

High level evolution:

- visualization improvements

- integration with existing vis. frameworks

- easier for non-expert users  (configuration, use, analysis)

- broaden focus to include LAN performance 

- Cloud is important! Make sure pS works well for cloud use cases.

- Simplifying configuration

- Stability improvements

- Global analysis (not just my mesh, but a view into results around the globe of other operators’ pS tests)

- Better training / documentation is desired. 

Attachment: perfSONAR User Survey Summary.pdf
Description: Adobe PDF document

Attachment: smime.p7s
Description: S/MIME cryptographic signature