Ron Bostick is an IBM'er who is heavily involved in the OSI standards process. He was just returning from a trip to China, and did not have an opportunity to provide handouts for this talk. I recorded five full pages of notes during the hour, trying desperately to make sense of it all. It was probably the best session I went to all week.
The OSI model defines protocols for communications between systems of any make. It encompasses everything from the logical interface with the application program to the physical connection (what kind of cable and its electrical characteristics). It is a sort of SNA superset.
The OSI model is built up in seven distinct layers. Each layer provides a specific service, and communicates with immediately adjacent layers via a strictly defined protocol. The layers are numbered 1-7 and are:
Intermediate nodes can be involved in passing a message between two endpoints, but these do not involve any but layers 1-3. When an application passes a message to layer 7 for transmission to another application, layers 7 through 1 all add some overhead to the original message. The header information added by each layer is stripped off by corresponding layers at the other end of the connection.
Some time was spent in talking about the X.400 and X.500 standards. X.400 is a store-and-forward message handling system (MHS) that implements both layers 6 and 7 (i.e. it sits on top of the session layer). You can use an X.400 mail system to compose and send a message to a user at another X.400 mail site. Your computer does not have to be connected to the other computer at the time you hand off your message to the X.400 program; the message is stored and forwarded to intermediate X.400 nodes, which store the message in turn until a link is established with an appropriate end node. Eventually the message gets to its addressee. (If your network is all leased lines, you don't need X.400.)
X.400 implements something called a "probe". If you don't know the name of your correspondent's computer, X.400 will broadcast a "probe" message on your behalf. A probe message is sent all through the network, and when an X.400 system out there recognizes the addressee, it responds with its system name. Probes are rather expensive and time-consuming, and are being replaced with the X.500 directory standard.
X.500 is "the baby" on the block; it is a forming standard. It defines a directory of users for the network. It currently allows a single user to appear in only one directory, but this is being reviewed. It is possible today to search a specific directory on a specific machine, or to broadcast a request to search ALL directories on the network for a certain user. Sounds expensive to me.
FTAM is the "File Transfer Access Method" and is a layer 7 implementation. You can use FTAM to:
This is all hairy stuff.