Why is error control needed at every OSI layer?

Question

I read that error control occurs at all seven layers of the OSI model. Why is error control needed at every OSI layer? What would be the consequence of only having one layer responsible for error control? Also, if error control only occurred at one layer, then what layer should that be?

Answer 1

The OSI model is a theoretical model, and there is nothing in the real world that strictly follows it. In fact, OSes have never implemented separate layers 5 to 7.

Even with the layers that you can align with some implementations in the real world, not all layers always implement error control.

• Many layer-1 protocols do not even have error checking.
• Many, but not all, layer-2 protocols have an error checking, and most that do simply drop damaged frames.
• Some layer-3 protocols, e.g. IPv4 (header CRC, but nothing for the payload, simply drops damaged packets), implement some type of error checking, but others, e.g. IPv6, do not.
• At layer-4, some protocols, e.g. TCP have error checking and correction, but other may not (UDP CRCs for error checking, but not correction, are optional for IPv4, but mandatory for IPv6).
• Some, but not all, applications have error checking.