Skip to main content

intro-networking

LANs to the internet​

OSI model​

Intro​

why OSI?

the OSI model is crucial for anyone working with networks, as it provides a common language and reference point for discussing network functionality, troubleshooting issues, and designing network architectures.

It helps to visualize the flow of data and the protocols involved at each stage, making network behavior more predictable and manageable.

Here is the basic data flow of the layers:

  • Layer 7: request/response abstraction over segments, establish a session connection.
  • Layer 4: segments, establish packet sequence and port destination
    • Add destination port and source port.
  • Layer 3: Packets, establish IP address source and destination info
    • Add destination IP address and source IP address
  • Layer 2: frames, establish MAC address destination
    • Add destination MAC address and source MAC address
  • Layer 1: electricity or light

application layer

The upper layers are layers 7 to 5, and are concerned with software and data preparation.

  • Layer 7: Application Layer 🌐 – The layer closest to you. It includes network-connected applications like web browsers using HTTP.
  • Layer 6: Presentation Layer 🔏 – Translates data into a readable format, handling things like encryption, decryption, and data compression.
  • Layer 5: Session Layer 🤝 – Manages the dialogue between two devices, opening, maintaining, and closing the connection session

network layer

The lower layers are layers 4 to 1, concerned with the actual transmission of data and connection of devices, moving information from point A to point B.

  • Layer 4: Transport Layer 📦 – Manages end-to-end communication and data reliability. It chops up large data into smaller pieces (segmentation) using protocols like TCP (reliable) and UDP (fast).
    • Core idea: Uses TCP or UDP and port numbers to specify data flow and process destination.
  • Layer 3: Network Layer 🗺️ – Handles routing across different networks. This layer uses IP addresses to guide data to its final destination across the internet.
    • Core idea: Establishes source and destination IP addresses, sending data from the default gateway (your router’s IP address) to the destination IP
    • core component: router
  • Layer 2: Data Link Layer ✉️ – Handles local node-to-node communication. It uses hardware-based MAC addresses to deliver data to the correct device on your immediate, local network.
    • Core idea: Establishes source and destination MAC addresses, sending data from the NIC to the router.
    • core component: switch and NIC
  • Layer 1: Physical Layer 🔌 – The actual hardware layer. It deals with cables (like Cat 6), Wi-Fi radios, and transmitting raw data bits as electrical or light signals.

Here is the flow going from top to bottom

  1. layer 5,6,7: YouTube and your IP address open an HTTP session.
  2. layer 4: you send data packets via port 80 using TCP
  3. layer 3: You want to send data from your default gateway (the source IP, which is the IP of your router) to the destination IP (YouTube)
  4. Layer 2: The NIC sends ethernet frames to the switch with the destination being the default gateway, which is just the router of the LAN.

what is OSI?

The OSI (open systems interface) model has 7 layers, where data flows from bottom (last layer) to the top (first layer)

  1. physical devices: The Physical Layer is responsible for the physical transmission of data over a communication channel. It defines the physical characteristics of the network, such as the voltage levels, data rates, and physical connectors.
    • Examples: Devices that transfer binary data to a network, which includes physical cables, and switches. Wireless wifi signals live at this level
  2. data link layer: Responsible for creating reliable connections between devices on a network. Organizes and breaks down packets that come from the third layer into smaller pieces called frames
    • MAC addresses are a unique layer 2 address given to every NIC.
  3. network layer: Deals with addressing and routing (the router is a layer 3 device), and tries to find the best path for packets to travel, which is called routing
  4. transport layer: Deals with packet transfer and formatting data into packets, ensuring that packets are transported reliably and does error handling to ensure packets are delivered in the correct order
    • protocols like TCP and UDP live on this layer
  5. session layer: Responsible for opening, maintaining, and closing a connection between two or more devices. Think of SSH.
    • This level establishes a session tag for those byte streams. This way, we know who is sending data to our server through that session tag. For example, if multiple requests come through, we know that the data belongs to a specific session.
    • Examples: SQL, cookies + sessions, RPC (remote procedure control)
  6. presentation layer: Responsible for transforming digital data into data computers can read, using techniques such as decryption and decompression to achieve that. The presentation layer is responsible for translating, encrypting, and compressing data. It ensures that data is presented in a format that both the sending and receiving applications can understand.
    • Examples: SSL/TLS, Data compression like GZIP, Text format conversion

example

Here is an example of the full OSI flow:

  1. Application Layer: You open your web browser and type in a URL (e.g., www.example.com). The HTTP protocol at the application layer initiates a request to the web server.
  2. Presentation Layer: The presentation layer encrypts the data using SSL/TLS to ensure secure communication between your browser and the web server.
  3. Session Layer: The session layer establishes a connection with the web server, managing the session and ensuring that data is properly synchronized.
  4. Transport Layer: The transport layer (using TCP) segments the HTTP request into packets, adds sequence numbers, and ensures reliable delivery to the web server.
  5. Network Layer: The network layer adds IP addresses to the packets, routing them across the internet to the web server.
  6. Data Link Layer: The data link layer adds MAC addresses to the frames, transmitting them over the local network to the next hop (e.g., your router).
  7. Physical Layer: The physical layer transmits the data as electrical signals over the network cable or as radio waves over Wi-Fi.

OSI layer purposes and headers

Before even sending a request, all the layers work together to tack on metadata that will be sent as request network headers on the individual packets being transmitted.

Here is how the request content is formed, going in order:

  1. layer 5,6,7: The actual payload content gets added to the request
    • layer 7 header added: which application layer protocol to use
  2. TCP/IP (layer 3, 4): Deals with describing the data flow between the source IP and destination IP and mapping out the path to reach the destination IP address.
    • layer 4 header added: Based on the layer 7 protocol defined, either uses TCP or UDP. Also decides which port to use.
    • layer 3 header added: defines source IP address (IP address of default gateway) and destination IP address
  3. layer 2: Deals with routing the source MAC address to the default gateway and how the router of the destination IP directs the connection to the destination MAC address.
    • layer 2 header added: source MAC address and destination MAC address
  4. layer 1: The physical connections in the source LAN and the destination LAN.

Layer 2​

Layer 2 (Data Link Layer) uses MAC addresses to deliver the data received from layer 3 to the correct physical device (like your local router) on your immediate, local network via frames

Layer 2 includes these components:

  • MAC addresses of NIC interface
  • frames
  • Switch
  • Router

Layer 2 includes these protocols:

  • ARP

communication in depth

  1. The NIC sends ethernet frames to the switch with the destination being the default gateway, which is just the router of the LAN.
  2. The router receives the data and prepares for layer 3 communication

Layer 3​

Layer 3 (Network Layer) uses IP addresses to route your data across different networks and guide it across the internet to its final destination, which is the public-facing IP address of the LAN in which your device is connected to.

communication in depth

  1. You want to send some data to a specific IP address, so first you send that data via packets (layer 4 functionality) from the source IP address to the default gateway, which is the router of your LAN (layer 2 functionality)
    1. To send data from the source IP address to the default gateway, the layer 2 ARP broadcast executes to resolve the source IP to its associated MAC address
    2. Then it resolves the default gateway to the router’s MAC address and then sends ethernet frames of the layer 4 data packets from the source MAC to the destination MAC
  2. The router decides what hops to do and routes traffic through the hop list until it reaches the destination IP.

Layer 4​

Layer 4 is responsible for end-to-end communication and managing how data flows. It takes the big chunk of data from your application and chops it up into smaller, manageable pieces called packets.

To do this, it primarily uses two different protocols:

  • TCP (Transmission Control Protocol): Focuses on reliability. It checks to make sure every single piece of data arrives safely and in the correct order. If a piece gets lost, TCP asks for it to be sent again.
  • UDP (User Datagram Protocol): Focuses on speed. It sends the data as fast as possible without stopping to check if the receiver actually got it.
propertiesTCPUDP
connection?Maintains a connection so the client and server can perform error-checking, starting with a three-way handshakeconnectionless
error-checkingError checks to make sure that packets always arrive in the correct order and without corruptionqNo error-checking

The packets vary in their data depending on which protocol is being used:

  • TCP: packet contains IP address of sender, payload, and TCP segment.
  • UDP: packet contains IP address of sender, payload, and UDP datagram.

Aside from choosing between TCP and UDP, Layer 4 has one more critical job: it uses Port Numbers to host different application level protocols (forms structure of data) and then uses either TCP or UDP to control the data flow behavior.

  • Example: HTTP is a layer 7 protocol, but it uses port 80 by convention and TCP, which are layer 4 components.

port numbers​

Think of your computer's IP address like the street address of a large apartment building. It gets the data to the correct building (your computer). But once the data arrives, how does your computer know which specific room (application) the data belongs to?

Different port numbers correspond to different protocols, which correspond to different data flow and behavior.

  • 🚪 Port Numbers act like the apartment room numbers.
  • 🌐 For example, standard web traffic (HTTP) uses Port 80.

When you open a web page, Layer 4 tags the data with Port 80 or 443 so your computer knows to hand that data to your web browser, rather than your email app or a video game.

TCP in depth​

TCP is a connection-oriented protocol. Before it sends a single byte of data, it establishes a reliable connection with the receiver using a process called the Three-Way Handshake, which is used to initiate the start of the session.

[ Your Computer ]                      [ YouTube Server ]
       |                                       |
       | ------ SYN (Let's synchronize) -----> |
       |                                       |
       | <--- SYN-ACK (I'm ready, let's go) -- |
       |                                       |
       | ------ ACK (Got it! Sending data) --> |
       v                                       v

Once the connection is active, TCP manages reliability through several advanced features:

  1. 📑 Segmentation & Sequencing: TCP chops large application payloads into smaller pieces called segments. It gives each segment a sequence number so the receiving computer can reassemble them in the perfect, original order—even if they arrive out of sequence.
  2. ↩️ Acknowledgments (ACKs) & Retransmissions: The receiver/client must send an acknowledgment back for the data it gets. If the server doesn't receive an ACK (an acknowledgement that the client received the data) within a certain timeframe, it assumes the data packet was lost in transit and automatically sends it again to the client.
  3. 🚦 Flow Control: If a fast server overloads a slower device with too much data, the receiver can signal the server to slow down, preventing the network buffer from overflowing.

Each segment has a destination port number so TCP knows which running process on the specific device (specified by layer 2 and 3) to send data packets to.

To make this happen simultaneously, your computer uses a combination of identifiers called a socket. A socket pairs an IP address (Layer 3) with a Port number (Layer 4).

So under the hood, your computer sees two completely distinct destinations:

  • 🌐 192.168.1.50:80 (Your IP + Port 80) -> Handled by the regular HTTP web browser process.
  • 🔒 192.168.1.50:443 (Your IP + Port 443) -> Handled by the secure HTTPS web browser process
TCP segments​

A segment in TCP is the name for the data sent via that protocol, and is part of the overall packets being sent in the protocol.

Here are the components of the segment:

  • sequence: The sequence number of the packet, used for ordering packets so the data can be constructed in order.
  • checksum: A unique hash value calculated by inputting the segment sequence number and packet payload together. This is used for error checking.

TCP performs error-checking with two mechanisms:

  1. checking sequence: If the sequence numbers of packets received by client and server have a mismatch, then the server resends the packets that didn’t make it.
  2. checking checksum: The checksum ensures data integrity because if the checksums of the segments the server sends and client receives don’t match, that means the data is corrupted.

TCP ensures data integrity knowing when a packet is corrupted because the checksums of the client and server don’t match, when the server sends the packet and the client receives it. This can happen because of one reason:

  1. payload is corrupted: The client and server have different versions of the actual data, so the checksum is different.

NOTE

When a packet is corrupted, the client resends the packet.

Imagine you are using an application that relies on TCP, and a momentary glitch on the physical wire drops 3 out of 10 data segments being transmitted. Let’s walk through what happens to ensure complete data transmission:

  1. Spotting the Gap: Your computer sends 10 segments, each stamped with a sequence number (e.g., 1 through 10). If segments 4, 5, and 6 get dropped by the network glitch, the receiving server notices the gap because it receives 1, 2, 3, and then suddenly jumps to 7.
  2. The Request: The server sends an ACK back to your computer, essentially saying, "I safely got up to segment 3, but I am still waiting on segment 4."
  3. The Retransmission: Your computer realizes that the timer for segments 4, 5, and 6 ran out without receiving an ACK. It retransmits only those specific missing segments until the server acknowledges they have arrived safely.

UDP in depth​

UDP is a connectionless protocol. It does away with all the administrative overhead of handshakes, tracking, and acknowledgments to deliver the absolute fastest speed possible.

Here are the two main properties of UDP:

  • 📤 Fire and Forget: UDP simply slaps a destination port number onto the data payload and shoots it out across the network. It does not check if the receiving device is online or ready.
  • 🤷 No Guarantees: There are no sequence numbers. If data packets arrive out of order, or if some are dropped entirely along the way, UDP does not care or attempt to recover them.

Here are the use cases for UDP

  1. voice over IP: realtime talking, calls, etc.

terminology

  • datagrams: the name for data packets sent via UDP

Protocols​

Intro​

A protocol is a system that allows two parties to communicate, and designed with a set of standard properties.

NOTE

The core idea to understand is that protocols are designed to solve a problem, and then people adopt them as a universal standard for universal compatibility.

TCP was designed to solve a problem, but it's not enough today for our massive needs, so that's why we designed new protocols.

These are the core components behind protocols:

  • data format: whether the data will be text-based like JSON or XML or binary like protobuf or HTTP2.
  • transfer mode: whether the data will be sent via messages like in UDP or HTTP, or as a stream like TCP or WebRTC.
  • addressing system: how to address source and destination to know where to send data to. Components include DNS, IP, and MAC address
  • directionality: whether it's bidirectional, unidirectional.
  • state: whether it's stateful like gRPC or TCP, or stateless like UDP or HTTP

NOTE

The main problem behind TCP is that it's sent as a stream while HTTP is message-based. Since HTTP uses TCP it has to constantly parse the stream to find where a message starts and ends, which is part of the overhead of TCP being used with HTTP.

Network topologies​