Let's bring this back to something we can easily relate to, something we all know and use. When humans greet each other we say hello, we wave, we shake hands. It would be odd if we greeted each-other by throwing an apple, or whistling a show-tune, or doing a hand-stand. Every language and culture has a way to greet another person and a way to communicate with human language. And we learn at a very young age how to do this. It makes it easier for people to work together and share information.
With the machines we have built and the Internet of Things, we also have a need to communicate. So, we have a language, and guidelines. There are millions of computers, sensors, actuators, systems on the internet, and for all things to communicate we need a shared language and rule-set. A shared language allows quick and efficient cooperation between things, using less programming, power, memory and data transmission.
You may remember 7-layer OSI model (still not to be confused with the delicious 7-layer burrito) from the last episode. We mentioned that this design is very idealized, and that you can't really completely separate the layers, so transports is pretty much the bottom four layers:
Again, it's really specific to the transport - almost all have built in error correction and reliability management (Ethernet, WiFi, Cellular...) but some do not (mostly, radios) and so that part of the transport may be something you have to manage.
The next two layers, Session & Presentation plus maybe a bit of the Segments layer, is what we deal with in the Protocols section. That's where you'll actually package data up, encode or encrypt it so that you can present it to the Application layer which is the final product you're making.
Going back to our comparison of languages for humans as a way to understand the languages for computers, we can also use the OSI layer division when talking about human language:
The transport of language can be air - in which case the language will be vocal.
The transport of language can be light - in which case the language will be visual - this is what sign language uses.
The transport of language can also be paper - in which case the language will be written!
Either way, the transport varies but the language itself is the same: English (or Indonesian! or Spanish! etc.) You can always get the same data across no matter the transport. In this sense, the protocol (language) is transport-agnostic, just like that OSI layer model above.
The way we talk and the way we write does vary. You get intonation when using air, exactness with writing. ASL even uses the space around the speaker to indicate relationships and the movements to convey emotion, irony and puns. So sometimes the transport and protocol bleed into each other a bit.
Likewise, there are some transports that are really incompatible with some protocols - for example you really don't want to try to run HTTP over Bluetooth Classic, as it would be terribly slow. And SigFox won't work with CoAP due to the packet size restrictions.
We'll try to call out some of these differences, but be aware that that just because you may be able to interchange the layers in theory, it can be very difficult to maintain a good customer experience in practice.