This project simulates a load balancer test to calculte the cost of running services on multiple configuration. It's main goal is to find the lowest possible cost of running a given load of service by twiking it's configuration while testing various scenarios.
It reads data from a text file containing one interger number per line. The first two numbers are used for configuration. All others number are treated as new clients that requested a service and so they need to be allocated to a given server. New servers must be launched if all running server reached their maximun capacity.
The first configuration number is ttask which tells the app how many ticks are used per client request.
The second configuraiton number is umax which tells the maximum number of tasks (or clients requests)
a given server can handle. If no servers are running or all running servers are using all availables slots
new servers must be launched to accomodate the new tasks.
This project was implemented using Python 3.9. Some features used requires at least Python 3.8.
First install the requirements creating a virtual environment and stalling the packages in requirements.txt file:
virtualenv -p python3.9 .venv
source .venv/bin/activate
pip install -r requirements.txt
Run the tests with pytest:
pytest
Run the app using the sintaxe: python src/app <INPUT_FILE> [OUTPUT_FILE]
python src/app clients.txt
INPUT_FILE is a required parameter and should point to a text file with one integer per line.
OUTPUT_FILE is an optional parameter to where the result should be sent. If no output file is provided the result is sent to sys.stdout
The app generates a log file with details of each run: server launched or removed, tasks assigned to a server and removed from a server. Also any predicted error will be logged to this file. Unpredicted errors are printed to stdout with the exception back track sent to the log file.
A configuration file was create in python format. The possible configurable values are:
SERVER_COST = 1.0 # Server cost per tick
TTASK_MIN = 1 # Minimun value for ttask
TTASK_MAX = 10 # Maximun value for ttask
UMAX_MIN = 1 # Minimun value for umax
UMAX_MAX = 10 # Maximun value for umax
OVERWRITE_DEST_FILE = True # Defines if the out_file (if informed) can be orverwriten if it exists
If you don't have Python 3.8+ in you system but have docker installed you can run the application and tests using docker. At this point you can not set an output file if you running the app using container. The results are going to be printed to stdout and you should redirect the result to a file if necessary.
To build an image for later use run:
docker image build -t load_balancer .
The steps below assumes you built an image with the command above.
You can run the app with the default sample clients.txt included in this source code just typing:
docker container run --rm load_balancer
If you need capture the output to a file use redirection:
docker container run --rm load_balancer > results.txt
To run the app with another file that exists in the container image (the file was present in the project directory when the image was built):
docker container run --rm load_balancer <INPUT_FILE>
To pass an input file not in the image to the container at run time use volumes:
docker container run --rm -v /FULL/PATH/TO/MY/INPUT/FILE.TXT:/app/clients.txt load_balancer
This will "replace" the file /app/clients.txt with your new file. As the file app/clients.txt is the default parameter for the app you dont need to inform it.
To run the tests:
docker container run --rm -it --entrypoint pytest load_balancer