20160318 - Meeting minutes, Friday March 18th 2016 - Task force SML/SMP
Estimated: 15:00 to 16:00 CET.
Performed: 15:05 to 15:53 CET
Agenda
- Introduction
- Approval of previous meeting minutes
- Approval of the agenda
- Review of the actions from the previous meeting
- Feedback from Oasis Technical meeting - change request on SMP
- Integration of the SMP with OpenNCP
- Business analysis 0.9 (see attached file)
- Cache mechanism
- AOB
- Connectathon
- Next meeting
Location
- AdobeConnect:
http://ec-wacs.adobeconnect.com/openncp/
Room Passcode: markus.kalliola or michele.foucart
------------------------------------------------------------------------------------------------
If you have never attended an Adobe Connect meeting before:
Test your connection: http://ec-wacs.adobeconnect.com/common/help/en/support/meeting_test.htm
Get a quick overview: http://www.adobe.com/products/adobeconnect.html
Adobe, the Adobe logo, Acrobat and Adobe Connect are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States and/or other countries.
------------------------------------------------------------------------------------------------
Participants
- Introduction
- Approval of previous meeting minutes
- Approval of the agenda
- Review of the actions from the previous meeting
- Feedback from Oasis Technical meeting - change request on SMP
- Joao and Massi attended the meeting with DIGIT. 8 change requests were submitted
- 7 change requests will be solved quickly
- 1 change regarding the requirement on digital signature is still under discussion => There is a chance that Oasis come back with a solution that does not suits eHealthDSI needs, impacting OpenNCP
- One of the problem arises because we are using a centralized SMP while the model is done for distributed signature
- If Oasis does not come with a solution, not wanting to implement the eHealthDSI use case (central SMP) => we need to take a decision, changing SMP or change trust model; other solution would be to use an extension (but not the preferred one)
- We are expecting a feedback from the Oasis technical committee in 4 weeks
- Joao and Massi attended the meeting with DIGIT. 8 change requests were submitted
- Integration of the SMP with OpenNCP
- Business analysis 0.9 (see attached file)
- Yves received feedback from Joao - First 15 points are easy to solve, for the other points, Yves will come back to Joao before distributing an updated version to the Community
- Cache mechanism
- Last meeting: SMP/SML waiting a decision for today
- Massimiliano Masi:
- research made on different cache mechanism possibilities. Merged TSL extension and SMP to achieve an implementation
- 4 scenariis => See OpenNCP page for the detailed analysis and tests results https://openncp.atlassian.net/wiki/pages/viewpage.action?spaceKey=ncp&title=Cache+implementation+through+ConfigurationManager+refactoring
- Redis
- Linear increase
- Redis + LUA
- Memcache
=> Redis and Memcache behave exactly the same - Microcaching - very similar in performance to Redis. DB kept as it is (little change in actual config), value increases
- Difference is on the standard deviation: time to connect to the database
- Limitations of the test: use of Eclipse, the object get cached after some time. The standard deviation is expected to be higher
- Change proposal solution (hibernate) :
- the proposal needs to be better clarified, it needs metric, need to be better clarified
- This solution can be complex to implement too, conclusion about simplicity to implement is "subjective".
- Publish this solution on the same page to make a comparison
- Redis
- Cache mechanism is part of the overall architecture, not only related to SMP/SML
- Need metrics, performance...
- In SMP/SML: discussion is how to store.
- Adrien confirms that in SMP/SML there is no strong requirements on cache mechanism. Cache limits the number of request. How it is implemented inside is implementation specific.
- We could go for a very simple solution (spring cache, simple annotation...). But does not want to interfere in the discussion.
- We should compare the different scenarios in term of implementation effort, time, costs... this dimension should be included in the table
- DG SANTE will share next week the results of the stress/load tests
- It is crucial to use the same approach!
- Business analysis 0.9 (see attached file)
- AOB
- Connectathon
- Tests in scope: Generate TSL, SML record and scrutiny test => same test as in Lisbon (won't test the on the fly lo-up for the metadata => Need to deploy an SMP like DIGIT did last time for Lisbon? We'll need to send a request to DIGIT
- Yves cannot come to the Connectathon for contract administrative reasons
- Next meeting
- Cache discussion 11am CET
- SMP/SML task force 3pm CET
- Connectathon