Web Basics

Web Essentials

Transfer: Networking/ OS basics, HTTP Content: HTML, Encoding, Dynamic pages

Networking Basics

Before being sent, files are broken up into packets of data

  • Circuit Switching: All the packets are sent along the same route. Cons: easy to hack because you know the exact route the packets are going to follow Pros: Good for real time things like voice, call, video
  • Packet Switching: All the packets are sent along different routes. Downsides: It will take more time because the packets need to be reassembled at the end Pros: Makes better use of resources, if one node in network is destroyed can still work because doesn't depend on a singular route

Network Protocol Stack Model

Open System Interconnection (OSI) Reference Model

The layers are

  1. User interaction (HTTP, FTP, SMTP)
  2. Data representation (XML, cryptography)
  3. Dialogue management (?)
  4. Reliable end-to-end link (TCP)
  5. Routing via many nodes (IP)
  6. Physical addressing (Ethernet)
  7. Metal or RF representation (Wifi, Bluetooth)
  • Network Transport Protocols: IP, TCP and UDP
  • TCP and UDP are built on top of IP
  • IP is the route and addressing between the nodes. TCP/UDP are the protocols for how to transfer the data
  • TCP is reliable end-to-end network. It has error checking, makes sure packets get to recipient in the correct order
  • UDP is meant for speed. It uses IP directly, as a forwarding service. It's a one way connection, not reliable, and no error checking. Often used for streaming

Network Transfer & HTTP (Hypertext Transfer Protocol)

Language client uses to talk to the server

  • Browser(client) opens TCP connection to server, writes request, server responds, connection closed (if HTTP 1.0)
  • It's stateless: every request is brand new
  • HTTP 1.0 has a non persistent TCP connection, 1.1 has a persistent connection HTTP Structure:
  • Break URL into hostname and path
  • Contact host at port and send any data to server
  • Download result code, bytes, and send to HTML renderer to show page

HTTP Server Design Approaches

  1. Wait until HTTP request arrives, start server, serve request, kill server (super inefficient)
  2. Sit in a loop waiting for requests (can only have one request at a time)
  3. Have a bunch of processes sitting around waiting (high memory overhead)
  4. Best option: Processes with threadpools (threads lighter than processes, but lose some memory protection)

URL Encoding

DNS lookup translates server name into an IP address (the numbers that a computer uses to identify itself on a network) DNS (Domain Name System) is a database of network names and IP addresses on the internet

  • Content Encoding: HTML (Hypertext Markup Language)

More Networking Basics

  • IP address assigned using Dynamic Host Configuration Protocol (DHCP) and mapped to target Ethernet address with Address Resolution Protocol (ARP)
  • ARP can be subject to man-in-the-middle attacks
  • Sliding window algorithm Handles communication buffering between sender and receiver.
  • Helps receiver handle input by ensuring the sender never transmits more than the advertised window size. The sender buffers unACKed data, receiver buffers data that is out-of-order or not yet read
  • 3 Way Handshake: Sender sends sequencenum x, receiver sends sequencenum y, x+1, and ACK, sender sends ACK and y + 1

Congestion

Handling buffering in the network

  • Buffers in middle of network, like routers, can become overloaded because they're not part of window negotiation -Queue and drop-tail technique Drop the incoming packet if the queue is full
  • AIMD (Additive Increase Multiplicative Decrease) -An algorithm for cautious network load calculation
  • Upon losing a packet, decreases this number. Upon success, increases this number. Trying to always figure out what's a successful transfer rate

Summary:

  • TCP Flow Control: receiver window
  • TCP Congestion Control: congestion window (max number of bytes to have in transit)
  • TCP Window: min { congestion window, receiver window }