For the sake of simple discussion, one can assume the "Cloud Platform" I talk about refer to Google Cloud or Amazon WS.
Currently I have a web server that mainly handles HTTP requests. However there will be real-time features (for example a live chat or video streaming) as well and those features are handled via socket-based connections in another server.
To clarify, actually the case is a bit more complicated: the "HTTP server" and the "socket server" actually both consist of multiple servers.
Currently the infrastructure is assumed to be such that all HTTP requests that goes to those HTTP servers will come through a load balancer and share the same IP address.
The socket-based connection is assumed to have much higher intensity of requests coming in than the HTTP requests.
Now here comes my problem: I don't know whether the socket-based connections should go through the same load balancer or will have a different load balancer assuming I am using
a Cloud Platform for all my servers.
On one side clearly using one load balancer will reduce cost (as only one public ip address is needed to be purchased), and the load balancer provided by the Cloud Platform has pretty good spec as far as I know. However based on my research, many companies not only use a different load balancer, but they actually do the load balancing via their own software instead of just using the load balancer provided by the cloud service for the socket-based connections, and only the socket-based connections. I don't really understand what kind of aspect I need to take into consideration in order to make a decision on this issue.
