OSDev.org

I'm reading computer networks 5th edition, and I'm a bit confused about how it works

Do network adapters just send raw data (Meaning with no additional information in a header and such) or not?

I've read about all the information packed into a frame that ensures it is properly transmitted across, so, is this handled by and up to the software to implement?

Or could I just send it raw data packed however I'd like assuming the receiving computer can understand it (Ignoring all protocols that may or may not be on top of it)?

Check the OSI model, which specifies and explains exactly that in detail. This table specifically might be enlightening.

Check out video series about the subject.

abcdabcd wrote:Or could I just send it raw data packed however I'd like assuming the receiving computer can understand it (Ignoring all protocols that may or may not be on top of it)?

The others already pointed to resources on the other questions. As to this one: Unlikely. If you send data on Ethernet, it better be in Ethernet L2 format. If not, any switch or hub in the middle is going to misunderstand, and any receiving network card can filter out the packet if it has any errors, like having the wrong destination address, or having an ill-fitting CRC. You can work around these issues, but on the whole, there is little to be gained from this. Now, inside of Ethernet, you can sent whatever. That is the lowest layer I would recommend using for custom protocols. But for most things, sitting on top of UDP is going to be preferable.

On the lowest level they send electromagnetic signals through the wire. Regarding the data integrity check, this is the job of TCP/whatever the protocol is used.