Palladio Component Model/Replication
Adding features to model replication more easily has been discussed and attempted several times. This page collects a list of attempts and options.
- briefly discussed at PCM Development/Palladio Concall/Minutes 20140811#Replications in PCM
- more detail at Architectural Templates and 
1:n mapping of AssemblyContext to AllocationContext
- Suggestion to allow one AssemblyContext to refer to n AllocationContexts
- Full explanation at PCM Changelog#1:n mapping of AssemblyContext to AllocationContext (Anne)
- Supported by PerOpteryx, PCM2LQN and PCM2StoEx as of PCM 4.0 (and previous releases).
- This has been rejected in a mailing list discussion on August 5th and 9th, 2011. Short summary:
- Is already supported by completions (not anymore, though)
- the 1:1 OCl check is useful to see whether the allocation is as intended (Anne: could remain a min 1 check)
- effort to adjust all analyzers
- assumption that local communication is preferred (see above link) is shaky
- what about probes for the results?
- a load balancer is a design and not an allocation question (Anne disagrees)
Replication could be, should be, or was modelled as a completion as proposed in Lucia's diss. Not clear what the status is, but probably with the discontinuation of QVT-R it is quite outdated.
Keywords: Replication, Replikation, Replica