Your web server will spawn new workers for each http request until it reaches the limit. The limit of workers should be adjusted according to the number of CPU and the amount of RAM available.
Each worker has its own independent instance of the application and is largely not aware of other instances. Obviously database records and cache are shared between workers.
In reality, it is unlikely that you are handling thousands of requests 'at the same time', unless you have a server farm and a team of devops.