In some sites , It is given that tcp/ip has 4 layers,network access,internet,transport,application. But in some other references , it is given that it has 5 layers (instead of network access, physical and link). What is the standard actually??? Somebody knows well tell me pls.
It seems like a combination of circumstances here has led to some confusion. Just so everyone is clear, The TCP/IP Model has a slightly different set of layers when compared to the OSI Model.
To be clear, the OSI Model's top three layers - that is: Application, Presentation, and Session - are essentially collapsed into the Application layer in TCP/IP. Additionally, the bottom two layers - Physical and Data Link - are combined into the Network Access layer for TCP/IP.
Therefore, there are 4 layers in the TCP/IP Model. Specifically, they are the Network Access Layer, Internet Layer, Transport Layer, and Application Layer.
I believe you are conflating the DOD\DARPA stack for IP with the OSI model
see here for how the layers of each model map to each other.
keep in mind, there are aspects of protocol modeling that are purely conceptual, and don't necessarily mimic analog reality. for instance in modern networking, the osi L1 and L2 are implemented in the network card's circuitry and in its driver code, but there's no good way to tell where one ends and the other begins. afterall the task of sending a media-compliant frame across the network cannot be separated from the task of constructing a 802.3 frame with all the correct data structures.
The layers are not necessarily concrete (especially when viewed from app perspectives), and as such there may not be a "correct" answer. A protocol can have as many layers as as you choose to slice it up into, regardless of the architecture of the code and the circuitry.
Models are often just how you think about a problem, and in this case, both models are valid for different analytic purposes.
Not sure if this helps but.
The original TCP/IP Network Model started with 4 Layers
The second version of TCP/IP became 5 layers: changed the name of Internet Layer to Network layer and divided the link Layer to 2 layers
Daou, I. (n.d.). Why we should understand TCP/IP and OSI Models? Retrieved from CCNA Hub: http://www.ccnahub.com/ip-fundamentals/understanding-tcp-ip-and-osi-models/
It can be argued that there are 4 layers, even in the 5-layer model, as it seems that in the 5-layer model they are merely splitting the Link Layer into 2 layers: hardware and link.
The TCP/IP model (4-layer model) that is being talked about is the RFC 1122, which aligns with Cisco Academy: the important player here
reference: Internet protocol suite - Wikipedia
I was taught they were 4 in TCP/IP mapped as follows over "iso/osi" layers, example of protocols between square brackets:
ISO/OSI TCP/IP (extended) Data unit's name (ISO-OSI) Application -> Application [telnet] => Packet (APDU) presentation -> Application => Packet (PPDU) Session -> Application (Security) [TLS]=> Packet (SPDU) Transport -> Transport [TCP/UDP] => Segment (TPDU) ----------Above this line the communication is end to end------------- Network -> Network [IP] => Datagram (Packet) DataLink -> Host To Network => Frame (Frame) Physical -> Host To Network (Bit)
Answering your question the TCP/IP model has 5 layers: Application, tansport, network and host-to-network. But I'd add the 6th layer as now more and more common TLS security because it has it's own identifiers and is a layer between TCP and application.
I can remember this by finding the "identifiers" for each data unit relatively to more used connections:
Application has application defined ID (TLS has [implicit] message sequence number) TCP has segment sequence number Network has IP address H2N has MAC address
This makes, in my opinion, HTTPS as an example of "session" level (I'd call it "security" level because there is where most security operations could be bundeled) because each TLS connection has its own identifiers and is incapsulated into TCP segments.