Introduction

This project developed to run REST API with two POST endpoints:

• /append:
  • post data:

    {
        "poolId": 123546,
        "poolValues": [1,7,2,6]
    }
    

  • expected response: show the status values are inserted (poolId not exist) or appended.

    {"status": "inserted"}
    or
    {"status": "appended"}
    

• /query:
  • post data:

    {
        "poolId": 123546,
        "percentile": 50
    }
    

  • expected response: show total pool values count (length) and its calculated quantile (result).

    {
       "length": 4,
       "result": 4.0
    }
    

Required setup

Please use python version 3.6 or higher, and ensure below 2 additional modules to be installed (not included in default python package).

• requests==2.25.1
• pytest==6.2.2

How to run

• Run the server: python server.py, default host access: http://127.0.0.1:8081.
• Run pytest for the unit test: pytest test.py

Potential improvement

• There is a potential of loosing data becaused of httpd issues need to restart, we can set a frequently dumping pool_data to file on disk for backup. Once it recovered, it will reloaded the file to memory back.
• There is a potential of out of memory error as all values stored on memory at pool_data variable. We can consider reduce this risk by settingup a separate monitoring and alert at specifict threshold, so engineer can timely fix.
• In case, the system needs to be scaled and more available, we may need to consider running /append & /query task on multiple worker nodes. This would need load balancer to distribute task between nodes, message queue handle concurent /append posts, and central storage for commond access by multiple nodes.