Hello everyone!
I recently received an email from a learner who is studying for his CCNA Routing-and-Switching Certification and he had a few excellent questions about the OSI model and how, exactly data moves from one-layer to the next. I figured my response might prove valuable to others studying for their CCNA so...here it is!
.....

1. Learner-Question: In video of the osi model, you said that the session layer should provide the source and destination port number but the fields of those ports are at the transport header- my question is how does the session layer put this number on field which does not exist in that time (when i send the date the encapsulation process goes down from the app layer)?
   

   In order to thoroughly answer all of your questions below, one really needs to know about computer programming, APIs, etc…which frankly, I know very little about. But what I do know, I’ll try to explain. From my understanding, there are some kind of software “links” or “hooks” which are used to allow a program at one layer of the OSI model to communicate with a program at another layer. Many applications have software built-in that provide multi-layer functionality. For example, imagine that you open some kind of Terminal Client (like Hyperterminal, SecureCRT, PuTTY, etc). That software you’ve opened technically does not reside at ANY of the OSI layers. That software just provides the graphical display such as the buttons you can press, the pulldown menus available, etc. Now imagine that within PuTTY or Hyperterminal you press a button to initiate a Telnet connection. At that moment, the PuTTY software informs your CPU that the CPU must start the Telnet program. PuTTY provides the interface so you can see…and control…what is going on, but PuTTY itself is NOT Telnet. It’s simply the user-interface so you can control Telnet.
   

   The functionality of Telnet actually is actually composed of an Application-Layer process as well as a Session layer process, all rolled into one. At the Application layer, the Telnet protocol answers such questions as, “what is a “username” and what is a “password” and is that required? Shall it send data downstream to lower-levels of the OSI model one-bit-at-a-time or several bytes-at-a-time? How is the user supposed to know when Telnet is waiting for input, versus currently transmitting output?” etc etc. The Session-Layer component of Telnet knows that it should be “listening” for incoming sessions on port-23. And when initiating outgoing sessions, it should use a destination port-23. At some point, the Telnet protocol creates a hook (I think these are called APIs) that allows it to invoke the Transmission Control Protocol (TCP). TCP knows that as part of the datastructure it creates, it must reserve 2-bytes for a “destination port-number” field and another 2-bytes for a “source port-number field” but what TCP DOESN’T know is what numbers to place in those fields. So this API (or whatever it is) allows the Session-Layer component of Telnet to convey to TCP that it place the value of “23” in either the Source or Destination Port Number field (depending on who is initiating the Telnet session).
   

   You may now be thinking, “but what about the Presentation Layer? You didn’t include that in the Telnet process?”. I believe that once SecureCRT (or PuTTY or Hyperterminal) invoke your Application-Layer protocol (such as Telnet or SSH) that SecureCRT/Hyperterminal will provide the Presentation Layer-component. SecureCRT knows if, when you press a keystroke on your keyboard, that key should be represented by ASCII or EBCDIC, SecureCRT/Hyperterminal also knows if you pressed the button indicating that encryption should be used. So it kind of “merges” or “blends” all of that information into Telnet thus providing the Presentation-Layer components. I’m not sure HOW it does this…but it does.
   

2. This question is about the type code field which lays at the llc sublayer, I understood that it purpose is to provide the upper layers what protocol is "talking"', how does it happens if the nic strips off the frame header in the decapsulation process?
   

   Basically what I wrote above happens in reverse here. There is some kind of internal software “hook” (probably another API) that allows your Layer-2 protocol (Ethernet) to communicate the value in the EtherType field to the CPU. In this way the CPU knows if it needs to invoke a Layer-3 procoess (like IP) or…if that process is already running…to take the Data from the Telnet frame and forward it to the correct layer-3 process. So IP itself does NOT see that Ethernet frame or any of the fields within it. But that “hook” (API???) provides the interface so that Ethernet data can be transferred upstream to the IPv4 process. At this point, my knowledge of the specific details of how this works ends.

3. If the type code provides the protocol(and its version), why does the IP header has "vers" field?
   

   Once again, to answer this question I believe it’s all about the APIs that allow protocols at different layers to talk to each other. Moving downstream (from Layer-3 to Layer-2) when IPv4 (as an example) has created a full IP Packet, it will “call” the API that allows it to hook into the Layer-2 protocol. IP doesn’t even CARE what that Layer-2 protocol is. It probably does something like, “Hey Layer-2 hooking API!! I’ve got some data here. Please pass it on to whatever protocol is operating at the Datalink Layer for me!!” The API, because it is talking to IPv4 will then invoke whatever layer-2 protocol is running (Ethernet, HDLC, Frame-Relay, etc) and say, “I’ve got some Layer-3 data for you!!”. At that point, the Layer-2 protocol (Ethernet in this case) will say, “Great! Can you give me some number that I can shove into my Ethertype field that indicates WHICH Layer-3 protocol created the data?? I don’t really care personally…but the device at the other end of the link receiving this data will need to know!”. So the API (that was originally called by the IPv4 process and was DESIGNED to be an interpreter between IPv4 and Ethernet) will say, “sure…the number you need is 0x800!” and thus…Ethernet places that value into the Ethertype field. Receiving an Ethernet frame would work the same way but in reverse. This time the Layer-2 protocol would “call” that L2-to-L3 API and provide the data, ALONG WITH the value of the Ethertype field to that API. In turn, the API would then know it needs to call-out to IPv4 and transfer the data upstream.