Myujth

I am developing a Python application which sends data to a web server via their REST API. The server uses digest authentication for the API calls. The theory behind digest authentication is to authenticate client in two steps. The initial hand shake uses a token, client receives WWW-Authenticate header from server and a new nonce and nonce count gets created. Then the second call is made with the actual API request with real data.

The original design is to keep HTTP client alive once the initial connection is established to have the client increment the nonce count with every request during the lifetime of the nonce. This will avoid the overhead of creating a thread and re-instatiating an HTTP client every time we need to send data to the server. The application is running as a service on a EC2 linux instance in AWS. So it is totally fine to keep the HTTP client alive.

import requests

sess = requests.Session()
sess.auth = HTTPDigestAuth(USERNAME, TOKEN)
sess.verify = False
sess.mount('https://', HTTPAdapter(max_retries=HTTP_CLIENT_RETRY_MAX))

However for next phase we are thinking going serverless, Lambda function to be specific. It scales much better and we don't need to maintain the EC2 server. The problem is every time Lambda function runs to send data to server, it has no memory of the previous API call. That means it will re-instantiating an HTTP client every time which is not efficient. I wonder if there is a way to keep a HTTP client alive using Lambda function.