Subject: RE: TOSCA Service Template Plans
My apologies for the delay, but I acknowledge both your comment emails of March 4th. While I cannot guarantee that the TC will choose to address the enclosed comments/questions, I will post these as issues in the TC issue tracking system for future TC consideration.
In addition to the below, some more comments:
1. How should one configure an initial instance count? For example, start with 3 VM that host 3 App Servers nodes (even when there is no scaling policy at all).
2. When linking two software nodes (MySQL, Monit) to the same Compute node (VM) – does it mean that the engine can choose whether to place them on the same VM instance or to place them on separate VMs that were created based on the same Compute node? How can a user choose?
3. When node A uses get_property(B, ip) – How the engine should select the correct instance of B? It is simple if A and B ratios are 1:1 or N:1 but it is unclear if the ratio is N:M.
We are going to develop a TOSCA engine as part of an applications orchestrator service.
While diving into the details we’ve seen that the Plan section is defined in high level at the XML TOSCA version and doesn’t appears in the simple Profile YAML version.
Few question about that:
1. What is expected from a TOSCA engine implementation when a Plan is not presented in the service template DSL?
a. Should the engine understand the “plan” (e.g. the order to execute node interfaces action) from the service template topology and dependency graph?
2. Is the Plan expected to be added to the simple profile YAML version sometime?
3. Can you please point me to the latest stuff related to TOSCA Plans?