What is a packet?

So, to start off with, we should engage in a bit of nit-picking. The term I've been using throughout this chapter-packet-is not, strictly speaking, the most accurate term for what I've been describing. Until now, I've been using the word packet to describe the most atomic unit of data transmission over a network. However, for the sake of accuracy, I should note that the term packet refers specifically to the most atomic piece of data transmitted by the network layer of the Open Systems Interconnection (OSI) network stack. In the transport layer, where we'll be most concerned with it (since that is the lowest layer in the stack we'll directly interact with through C#), the atomic unit of data transmission is actually called a datagram. However, I'll note that it is much more common to refer to data units of the transmission layer as packets than datagrams and so will continue with this colloquial use of the term throughout the chapter and throughout the rest of the book. I did, however, want to take the opportunity to point out the distinction between the two terms in case you encountered either being used elsewhere in different contexts. With that in mind, what exactly is a datagram, or packet?

We already know quite a bit about what a packet must be in order to be useful, so let's formalize it into a definition. A packet is an atomic unit of data, encapsulated with sufficient context for reliable transmission over an arbitrary network implementation.

So basically, it's a payload (unit of data) with a header (sufficient context). This shouldn't be surprising by this point, but let's look at how this translates to an actual array of bytes passed from our transport layer to the network layer. To do so, we'll use Wireshark to examine the actual data packets being sent to and from my own Ethernet port, and look at how each part of that definition translates to actual datagrams.