[TC_28] NebulOuS manages the execution of IoT data processing pipelines

[Christos-Alexandros Sarros]

Objective:

Test if user can express an IoT data processing pipeline as a NebulOuS application and NebulOuS efficiently manages the execution of the pipeline.

Preconditions:

1. NebulOuS core is running.

2. Three NebulOuS nodes are active and registered in NebulOuS.

3. Application of “IoT APP” is registered in NebulOuS. The APP SLO is configured such as max_age for “step1_output” can’t be bigger than 30 seconds. The objective function is set to minimize the number of instances of step 2.

Steps:

Step 1.
- Action: User logs in into NebulOuS web UI and launches an instance of the application “Scaling APP”.
- Expected Result: NebulOuS deploys an instance of the APP step “step1”. No instance of step2 is created.

Step 2.
- Action: User publishes a message to the pub/sub queue “stream_input” of the application, with the body {“group”:”g1”,”t”:10}
- Expected Result: NebulOuS detects that the max_age of “step1_output” queue is bigger than 30 and creates an instance of “step2”. Pending messages are processed and max_age for the queue goes to 0. NebulOuS stops the execution of “Step 2” instance.

Step 3.
- Action: User publishes a burst of 500 messages to the pub/sub queue “stream_input” of the application, with a fixed value for “t” (10) and different values for “group” (“g1”,”g2”,”g3”)
- Expected Result: NebulOuS detects that the max_age of “step1_output” queue is bigger than 30 and creates an instance of “step2”. After a short period of time, NebulOuS reevaluates the SLO and confirms that the max_age is still bigger than 30 seconds and creates more instances of the “step2”. This goes on until all the pending messages are processed and max_age for the queue goes to 0. NebulOuS stops the execution of “Step 2” instances.