Date

Attendees

Apologies

 

Goals

  • Review the eduKEEP project plan
  • Discuss the tastks and the assignments
  • Check assignees availability for next immediate actions and tasks

Discussion items

TimeItemWhoNotes
eduKEEP project planDavide Vaghetti

 Davide shared the workplan draft.

35mintask "Develop ansible playbooks for architectures deployment (toolkit)"

Rolf Brugger

Davide Vaghetti

Justin Knight

Rolf queried if it's worth the effort on development of Ansible playbooks and if this was specific to the IdP.

Davide noted that in order for the pilot to run a means of deploying the infrastructure is required to avoid barriers in adoption (making it as easy as possible for participants to adopt). The Ansible would be for the IdP and core components. A mock attribute authority could also be created.

Rolf: Would Ansible playbooks be made for all four architectures identified?

Davide suggests just 2: eduID (Central IdP) and Account Linking/Proxy based.

Rolf warns it would be difficult to make a playbook for the eduID, and would need buy-in from SWITCH. The service is distributed over several hosts and is quite specific to the SWITCH use-case; difficult to make generic.

Davide: in the CampusIDP task an Ansible playbook is being created to deploy Shibboleth IDPs, we can also leverage existing solutions from many NRENS (GARR, CESNET, RENATER at least).

Rolf: for the pilot, could alternatively develop a single instance that pilot participants could connect their own infrastructures to, rather than each participant deploying an infrastructure at their end.

Davide: the single instance solution in that case should be multi-tenant.

Justin: should we develop just a reference architecture? Or aim more ambitiously with the playbooks that pilot participants can use? The latter helps to remove barriers to adoption and demonstrates proving the concept, but is a higher workload.

Davide: we have also another option, instead of producing playbooks, the task could offer just the reference architecture and assistance from task members for pilot participants to deploy it, but probably the effort would be bigger.

Key next step is to nail down and all agree on the scope of the task. Rolf will have a talk with Ann in the first case

15minOther comments on the plan

Sclifos Constantin

Justin Knight

Hervé BOURGAULT

Constantin: should we also specify migration strategies to the user-driven approach? (for the pilot)

Justin: try to propose migration in the pilot would be unbearable

Hervé: eduKEEP concept itself has to be better defined, use cases can help


Action items