Well, that is pretty much how it goes. Recover the ID, parse the rest of the message as the id indicates, ALWAYS testing if the data is in the right format since a hacker can send fake packages.

Also, most packet protocols I have seen use binary information - not text - since it this will save a lot of CPU and bandwidth (think that you need to send a character position at 1578.4783892301 is a 4 bytes float/8 bytes double, but 16 bytes string), not to mention conversion problems.

Unfortunately I don't know how this goes on C#, I never worked with it, I can tell you how to do so in python or C.

If you are new to networking, I would tell you to first get a simple client/server application working. In a very high level, it goes like this (the names in bold are the common functions/methods/class names, so you can google how to use them on your favorite language):

Server side:

- Create a socket, mostly with socket function or class.

- Use a bind function or method to reserve a port to that socket.

- Set the socket to listen state.

- Wait in a loop with accept. Accept will likely to be a blocking operation, so your loop will stay blocked in it, consuming no CPU, until a new connection arrives.

- When a new connection arrives, accept will give a new socket, that will manage the server connection with the client.

- You then call recv to receive data from the new socket.

- The received data needed to be parsed.

- You can send data to the client by calling send.

- When data transfer is finished call close to tell you OS that socket is no longer needed.

Client side:

- Create a socket.

- Call connect with the server information.

- Call send and recv to communicate with the server.

- When data transfer is finished call close to tell you OS that socket is no longer needed.

With this simple model you should be able to make a very simple client server application that sends/receives a string. When you get it working you can try something more complex, but I would advise you to start small :D

Text mode marshaling is not very efficient. Typically, you will send byte buffers. Strings in C# (and Java, JavaScript, and may other dynamic languages) are inefficient if you keep changing/splitting/concatenating them.

Even more typically, you'll pre-allocate your buffers as arrays of bytes, and re-use them between operations (with async operations, you have to keep track of which are currently "in use" or not.)

Your marshaling and demarshaling code would typically take a triple of byte array, offset to start of data, and length of data block.

Often, this triple is wrapped in a writable stream subclass.