User Story 2-4: Reactor State Updates

[OmerMajNition]

Once deployment is running, and after some retrospection of the performance results, users may want to make state changes to the already running deployment without disrupting the services. For example, we may want to shift any cache of a server to a newer eviction algorithm, or change its partition ratio, or make changes to load balancing selection algorithm. Some parts of the day it makes more sense to allocate more cache for writes and other parts of the day workload may be read heavy, so changing read and write cache ratio would make more sense. Users could leverage this hook inside their production environment where they can seamlessly change the component states. Similarly, users could decide on switching to an already supported cache eviction algorithm, or business logic inside load balancer could be varied based on traffic analysis. In summary, we should let the users have a knob to steer the deployment on the go.

Implementation Considerations:
“Admin” like reactor can delegate the state update task to reactors and via mutation reaction each reactor could update states locally.