The TOO runtime model is built on software components that run independently and communicate seamlessly. Each component runs in a separate process with a dedicated service that automatically restarts it in case of failure. Components can also be distributed across multiple machines and locations—for example, a sensor programmed in C++ might be deployed inside a shrimp pool in Guayaquil, Ecuador, while another component runs in a cloud data center.

Consider the following three “things” in TOO:
sensor { }
switch { }
thermostat { sensor switch }

• The sensor has an instance named Guayaquil.
• The switch has an instance named Milan.
• The thermostat runs in the Cloud and includes both sensor[Guayaquil] and switch[Milan] as members.

In a component-based system like TOO, every real instance runs as a separate process—this means the thermostat process in the cloud does not create new processes for its members. Instead, these members exist as mock-up objects inside the thermostat.

A mock-up in TOO is a thin proxy that represents a remote instance without running a full process.

• The switch mock-up in the thermostat communicates with the real switch process in Milan to turn it on or off.
• The sensor mock-up interacts with the real sensor process in Guayaquil to retrieve its readings.
• All communication between the mock-ups and the real components happens via RPC (Remote Procedure Call) over MQTT.

This design ensures that each real instance runs only once, while other components interact with it remotely, making TOO an efficient and scalable system for distributed execution.