release2 03‐Using SeQaM - Smart-Edge-Lab/SeQaM GitHub Wiki

SeQaM offers multiple methods for users to interact with its functionalities. These methods enable the execution of commands and the management of experiments.

The primary method of interacting with SeQaM is through the web console, accessible via the web interface. Users can connect to SeQaM by accessing the console on port :8000 where the central components are deployed.

An experiment is a set of commands triggered at a certain point in time. Experiments are defined within the ExperimentConfig.json file, which describes a series of commands as part of the experiment. Once configured, the experiment can be executed through the experiment dispatcher as a console command.

start_module module:experiment_dispatcher

For more detailed information on configuring experiments, please read more on Configuration files here.

Additionally, users can utilize the on-demand events feature for interacting with SeQaM. If you need to create SeQaM events based on some custom events your algorithm generates (e.g., triggering a CPU load when an application has just migrated to a new server), you will need to send an on-demand POST request to the command translator module with the command you would like to execute.

curl -X POST http://Central_IP_Address:8001/translate \
     -H "Content-Type: text/plain" \
     --data "cpu_load src_device_type:ue src_device_name:ue003 cores:0 load:90 time:60s"

In this way, you can implement your logic to determine what command to execute in SeQaM and when.

A command represents an action to perform in any of the SeQaM components (central, distributed, network). The commands can be triggered from the console or automatically invoked by the experiment dispatcher module.

The general syntax of a command is

action arg1-name:arg1-value arg2-name:arg2-value ... argN-name:argN-value

where any or all argN-value can be optionally quoted with single or double quotes; these optional quotes become mandatory if a value contains a whitespace.

Generates CPU load in a desired device like user equipment or server.

Action: cpu_load

cpu_load time:duration load:load-value src_device_type:src_device_type src_device_name:src_device_name 
cores:number-of-cores comment:"any optional human-readable comment" mode:stress-mode random_seed:random-seed 
load_min:load-min load_max:load-max load_step:load-step time_step:time-step

Optional Arguments

📝 Note:

• The comment argument can help attach descriptive text to the experiment action.
• Ensure that the duration argument is suffixed with the appropriate time unit (s, m, h).
• Make sure the total time of the command matches the time-step x amount of steps when using dec or inc modes

Example

cpu_load src_device_type:ue src_device_name:ue001 cores:5 load:20 time:60s

Generates a defined amount of traffic in a specific link in your network scenario. This method uses a client and a server that must be attached to both sides of the link you want to stress with network traffic.

network_load time:duration load:load-value src_device_type:src_device_type src_device_name:src_device_name 
dst_device_type:dst_device_type dst_device_name:dst_device_name comment:"any optional human-readable comment" 
mode:stress-mode random_seed:random-seed load_min:load-min load_max:load-max load_step:load-step time_step:time-step

Action: network_load

Optional Arguments

Example

This syntax generates traffic from a source to a destination. Both devices should have implemented the network component.

network_load src_device_type:ue002 src_device_name:ue dst_device_type:server dst_device_name:srv102 load:50 time:5s

Action: memory_load

memory_load time:duration load:load-value src_device_type:src_device_type src_device_name:src_device_name 
workers:number-of-workers comment:"any optional human-readable comment" mode:stress-mode random_seed:random-seed 
load_min:load-min load_max:load-max load_step:load-step time_step:time-step

Optional Arguments

⚠️ Warning:

• Memory load, also will generate CPU load, due to the manner computers work.
• load argument in Memory load is in MB and not percentage

Example

memory_load src_device_type:ue src_device_name:ue001 workers:5 load:800 time:10s

This command launches the modules defined in the ModuleConfig.json file, initiating their respective operations.

start_module module:module_name

Action: start_module

Example

start_module module:experiment_dispatcher

⚠️ Warning: Only the experiment_dispatcher module is supported for now. The experiment_dispatcher module triggers the execution of experiments defined in the ExperimentConfig.json file.*

This command exits the modules defined in the ModuleConfig.json file, terminating their respective operations.

stop_module module:module_name

Action: stop_module

Example

stop_module module:experiment_dispatcher