Network Foundations Part 1: The Models

In the wonderful world of networking there are three foundational models that are used as a starting reference point for discussing networking interactions and functions. They are the OSI Model, the DoD Model, and the Cisco Hierarchical model. Each of these models has a slightly different focus, but all are important to know.

OSI Overview

The OSI model is the seven layer model that we typically frame device and protocol functions against. For example, you have probably heard of Layer 3 switching, that a hub operates at the physical layer, or that TCP is a Layer 4 protocol. These are all being framed against the OSI model. Below is a diagram of the OSI model.  Simple ways to remember the OSI model include from the top down “All People Seem To Need Data Processing” or from the bottom up “Please Do Not Throw Sausage Pizza Away.”

Physical Layer

I like foundations so let’s start from the bottom with the physical layer.  The physical layer is all about bits, and the media that we use to move those bits.  So if you have an ethernet cable, it is electrical signaling that makes bits, if you use fiber it is light, or if you use wireless it is radio waves; however, they are all ways to just move binary bits to another device or location.  We also need to decide who is going to decide how fast we go.  That is set by deciding who is the DCE (data communications equipment) and who is the DTE (data terminal equipment).  The DCE sets what we call the clock rate which determine the speed at which we can talk over a media.  In the modern Ethernet world this is hard coded on both ends or allowed to auto negotiate, but in test environments it is occasionally necessary to set the clock rate.  In the WAN and Internet worlds it is the service provider who establishes the clock rate.  Hubs are a device commonly found at Layer 1 as are CSU/DSUs (channel service unit/data service unit) for T1/PRI circuits and modems for cable/analog connections.

Data Link Layer

The data link layer is responsible for local physical transmission and error detection.  The data link layer is split into two functions or sublayers if you are using Ethernet, the Logical Link Control (LLC) and the Media Access Control (MAC) layers.  The LLC sublayer takes data from the network layer which is identified by protocol and packages it with flow control information and hands it off to the MAC sublayer.  The MAC sublayer has the physical addressing of a given device, decides how to place the frames onto the media, and provides basic error detection via a frame check sequence (FCS).  Basic Switches and Bridges operate at this layer.

Network Layer

The network layer is responsible for logical addressing and determining the path to reach other devices.  Layer 3 protocols are groups into routed protocols, like IP and IPX, and routing protocols, such as EIGRP, OSPF, and BGP.  The most common logical address is the IP address.  The IP address is a decimal (IPv4) or hex (IPv6) value that is either manual configured per device or automatically assigned via the use of DHCP (Dynamic Host Configuration Protocol).  There are also many other layer 3 protocols that have their own unique addressing schemes such as IPX and Appletalk.    The second function of the network layer, path determination, might be the golden key to the kingdom.  It ensures that we know how to locate any given device on the network by its logical address.  This process is also refered to as routing.  The network layer uses packets as it’s container of choice, and routers and layer 3 switches operate here.

The Transport Layer

The transport layer segments and multiplexes sessions together to provide end to end data transport services.  This data transport services are identified by a port number.  I’m sure you’ve heard of a TCP or UDP port.  Those end to end data transport services come in two flavors: connection-oriented (reliable) and connectionless.

Connection-oriented is referring to the fact that we agree to the transfer (handshake), maintain flow control to ensure all data that is sent is received, and tear down.  In the Cisco world, connection-oriented has a virtual circuit setup, sequencing, acknowledgment, and uses flow control.  Flow control is a combination of buffering (temporarily holding data before processing it), windowing (establishing how much data can be sent before an acknowledgment), and congestion avoidance.  We commonly use file sharing, web browsing, and ftp as an example of connection-oriented applications.  The transport layer uses segments to describe it’s data.

Connectionless is very simple I send data and hope you get it.  We use connectionless when the overhead of the connection is not worth it.  For example Voice over IP (VoIP), we don’t care if one small piece of the data doesn’t get there because odds are it might not be truly missed, and retransmission is not an option because by the time we found out we missed one and sent it again the topic of conversation could have changed.  Voice, Video, TFTP, DNS are all good example of connectionless applications.

The Session Layer

The session layer separates each applications data from one other by creating individual sessions that are either simplex (one way), half duplex (one way at a time), or full duplex (bidirectional simultaneously).  We refer to session layer data as a session.

The Presentation Layer

This layer is always a hard one for me to define.  It provides data translation and formatting.  So ASCII text, JPEG images, and PDFs are examples of presentation layer formats.  Basically it ensure that data on the application layer of one machine is readable when it arrives at the destination machine.

The Application Layer

This is where we interact with the computer.  Think about things like Firefox, Evernote, Outlook, or Skype.  The application layer is responsible for checking the reachability of our destination and ensuring we have resources to get there.

Encapsulation/De-encapsulation

The process of moving from the top layer of the OSI model to the bottom is called encapsulation.  We call it that because when we start with the application data which is formatted by the presentation layer.  The presentation layer hands that off to the session layer who puts session information around that data.  Then the transport layer puts connection or segment information around it.  The network layer wraps that segment with packet information and hands it to the data link layer where it is framed just before it is turned into bits and put on the wire.  De-encapsulation is the same process in reverse.  By the way we call those bits, frames, packets, and segments protocol data units (PDU).

The DoD Model

The Department of Defense (DoD) model is focused directly on the Transmission Control Protocol/Internet Protocol (TCP/IP) protocol suite.  Basically it is a simpler way to view the OSI model in the modern world.  It takes the Application, Presentation, and Session layers and merges them into an Application layer.  The Transport layer becomes the Host-To-Host layer.  The Network layer is renamed to the Internet layer.  Finally the Data Link and Physical Layer get smashed into the Network Access layer.

The Cisco Hierarchical Model

This model is used to represent the basic foundation for building scalable, reliable, and maintainable network.  It consists of three layers: core, distribution, and access.  This is not always physically broken out into these three layers, but the idea is that we divide the functions up into logical concepts.

The core layer is the brain of the network.  It is responsible for moving traffic as quickly as possible without sacrificing reliability.  Rules for the core part of the network are simple avoid segmentation, security, filtering, or single points of failure.

The distribution layer is where routing, filtering, security, and WAN connectivity occur.  The distribution layers basically decides if it should forward data to a server or other core device, deny it due to security or network policies, or route it to a different part of the network.  This is the central nervous system of the network.

The access layer is where users connect into the network.  This is where we start segmentation, begin apply network policies, and provide connectivity to the distribution network for access to other services.