Protocols that perform error recovery (typically via forward acknowledgement) used to be called "reliable". Examples include 802.2 (a data link layer protocol) and TCP (a transport layer protocol). Has this terminology fallen into disuse?
I ask because Cisco's Internetworking Technologies Handbook, Fourth Edition (2006) states, "The data link layer provides reliable transit of data across a physical network link." Frame relay and ATM are layer 2 protocols that do not provide error recovery.
Saying that the data link control layer is reliable is, of course, quite wrong as a general statement.
I could imagine that, whoever put this text together may just be ignorant. For example, the person may have copied the text from some description of the OSI 7 layers without a critical evaluation of whether or not it was correct in the new context.
You quoted 802.2 - strictly I think that has to be a particular flavour of 802.2 - and TCP as being reliable. They are reliable because they use acknowledgements in order to be sure that a unit of data has arrived, retry sending if necessary from retained copies of the data and notify the higher layer when it is evident that data are not getting through.
802.2 - whatever the proper description of the reliable version is - is an example of a data link control layer protocol. TCP, however, is an example of a transport layer protocol and, because it is reliable, it places no reliability requirements on the underlying network and data link control layers. Thus, where TCP is being used as a transport layer protocol, it is actually desirable, in terms of not imposing pointless overhead, to have an underlying data link control layer protocol which is *un*reliable.
Chris Mas> Protocols that perform error recovery (typically via forward