experimenting with user discovery algorithms for a social app in LantasHackathon on 30th January 2016 - ensuring stable & quick response for discovering users from among a user's phonebook contacts against an evergrowing directory of millions of app users in the application's database
See benchmarking results of the experiment by clicking on the link.
Install the following python packages first:
pip install pymongo
pip install flask
pip install flask_restful
install MongoDB and have it running on localhost:27017 - no initial databases needed to be created for this exercise
Configure number of users you want to simulate in the MongoDB server by altering the num_of_users_to_populate_in_server variable in the populate_users.py file
python populate_users.py
----- allow some time for the python script to populate the selected number of dummy users into the MongoDB server
python run_server.py
----- flask_restful app will run on localhost:5000
Copy entire contents of the file phonebook.json and make a POST request to http://localhost:5000/users/discover/ by pasting the copied contents into the raw message body of the POST request. ----- note the value of the time_taken field inside the JSON response.
You may change the cross-matching algorithm by choosing between nested_forloop_lookup() or hashmap_lookup() by uncommenting the function call written inside the crossmatch_app_users() function in the run_server.py file. Repeat the POST request to retry with the chosen algorithm.
- Python VirtualEnv - to maitain an isolated virtual environment for this python exercise
- POSTMan REST Client - a standalone chrome app to make standard POST requests