Networks 4
foNetworks 4
Socket
- = an endpoint of communication between 2 devices on a network
- Is an interface between application layer and transport layer
- Located between operating system (e.g. Windows, Linux) and application programs
- To access the services provided by the TCP/IP protocol suite, an application needs to use the instructions defined in the socket interface
- Type of socket
- Stream Socket (TCP): connection oriented, reliable, ordered, error-checked delivery
- Datagram Socket (UDP): fast, unreliable, connectionless communication
- Socket address consists of a combination of IP address and port no.
- Basic communication flow of socket
- https://www.keil.com/pack/doc/mw6/network/html/using_network_sockets_bsd.html
- Bind is to bind to the port no.
Client-Server architecture
- = a network design model in which multiple client devices communicate with and request services from a central server
- Note: in socket programming, run server first before running client
- Server
- = a computer or program that provides services or resources to clients over a network
- A server program is an infinite program: when it starts, it runs infinitely unless a problem arises or shuts down
- Waits for incoming requests from clients
- When a request arrives, it responds to the request
- E.g. web server, file server, database server, email server
- Client
- = a computer or program that initiates a request for services or resources from a server over a network
- A client program is finite: it is started by the user (or another application program) and terminates when the service is complete
- Opens the communication channel using the IP address of the remote host & the well-known port address of the specific server program running on that machine
- IP address leads you to the network interface
- Port number indicates which program to send the data to
- Can have multiple servers running on the same server, but diff. ports
- After a channel of communication is opened, the client sends its request and receives a response
- Although the request-response part may be repeated several times, the whole process is finite → eventually comes to an end
- E.g. web browser, email program
- Advantages
- Centralized resourced management
- Server holds and manages resources (files, databases, applications, etc)
- Allows for central control of data, security policies, backups, and user access
- Scalability
- New clients can be added easily without changing the system
- Server can be upgraded (add more resources: RAM or using load balancing) to serve more clients
- Improved security
- Enforce security policies from the server side
- E.g. user authentication, file permissions, monitoring network traffic
- Ease of maintenance
- Software updates, backups, and system configuration are done on the server
- Reduces the need to configure every client device
- Reliability and consistency
- Data stored on the server is kept consistent
- All clients access from 1 source which should be most up-to-date
- Remote access support
- Clients can access servers from anywhere over the internet, enabling remote work, cloud services, and web applications
- If 1 client is down, the server and other clients are not affected
- Disadvantages
- Single point of failure
- If server is down, all clients lose access to service (whole network down)
- High cost of setup and maintenance
- Servers are expensive and require skilled administrators
- Initiating infrastructure and licensing costs can be high
- Network congestion and overload
- Possibility of network congestion, server overload, slow down network
- Complexity
- Configuration and management of a secure and efficient server environment can be complex
- Network dependence
- Clients rely heavily on a stable network connection required
Peer-to-peer (P2P) architecture
- = a decentralized network model where each computer (peer) acts both as a client and server
- No central server ⇒ each peer can request and provide services or resources directly to other peers
- Key features
- All peers have equal role
- They can initiate (send a request) or response to a request
- Decentralized
- No central point of control
- Resources are distributed among peers
- Resource sharing
- Resources or services are shared among peers
- Autonomy
- Each peer manages its own resources and may be online or offline at any time
- E.g. BitTorrent peer-to-peer file sharing, blockchain networks
- Advantages
- Low cost
- No need for expensive server hardware or infrastructure
- Scalability
- New peers can join the network easily to contribute resources
- System can grow without major reconfiguration
- No single point of failure
- 1 peer fails does not affect the whole network ⇒ limited impact to network
- Resource sharing
- Peers can share files, processing power, or bandwidth directly with others
- Disadvantages
- Lack of centralized control
- No central authority to manage users, resources, and security
- Harder to enforce rules and monitor activity
- Security risks
- Each peer is responsible for its own security
- Malicious peers can spread malware or unauthorized content
- Data consistency issues
- Files in peers not guaranteed to be up-to-date
- Synchronization is difficult
- Performance
- Performance suffers if there are few online peers or peers have low bandwidth
[Application layer] Protocols
Dynamic Host Configuration Protocol (DHCP)
- Automatically assign IP addresses and other network configuration parameters to devices (clients) on a network
- IP address
- Subnet masks
- Default gateways
- DNS server addresses
- Lease duration
- Without DHCP, every device would need to be manually configured
- Error-prone
- Inefficient
- Unscalable
- How it works (DORA)
- Discover: client sends a broadcast to find available DHCP servers
- Offer: DHCP server replies with an offer including an IP address and configuration
- Request: client requests to use the offered IP address
- Acknowledge: server confirms and leases the IP address to the client
- DHCP server sends the address to the client for acknowledgement
- Uses UDP for sending a broadcast to find available DHCP servers
- Client device does not have an IP address ⇒ cannot use TCP as connection needs to be established
- Light-weight, minimal overhead
- Connectionless ⇒ no handshakes, no session setup
- Types of IP address allocation:
| Static IP address | Manual configuration: device manually configured with its IP address, subnet mask, gateway, DNS DHCP server is not involved ⇒ no DHCP request is sent DHCP reservation: using MAC address of the specific device Always get the same IP address from the DHCP pool ⇒ a DHCP request is required to obtain the IP address |
|---|---|
| Dynamic IP address | DHCP server leases a dynamic IP address (reusable) to the client for a period of time |
Domain Name System (DNS)
- Translates domain names (e.g. www.google.com; human-readable format) into IP address (used for routing; machine-readable format)
- User uses a file transfer client to access file transfer server at a remote host
- Only the server name (e.g. www.google.com) is known, but TCP/IP suite needs IP address of the server to make a connection
- Steps to map host name to IP address:
- User passes host name to file transfer client (FTC)
- FTC passes host name to DNS client
- DNS client sends a message to DNS server with a query containing the FT server name, using IP address of DNS server (pic on right)
- DNS server responds with IP address of FT server
- DNS client passes IP address to FT client
- FT client uses IP address to access FT server
- Hierarchy
- Name-address resolution steps:
- Check local cache (on host / DNS client)
- Client (resolver) checks its own cache
- If IP address is found locally, client uses IP address to route the message in the network ⇒ no need step 2
- Query Local DNS server (ISP’s DNS resolver)
- ISP = Internet Service Provider
- If not found locally, client sends query to the DNS server specified in the IP configuration
- Server checks its own cache
- Query the Global DNS hierarchy (if not cached)
- If ISP DNS resolver does not have the answer, it must contact the DNS hierarchy of distributed database
- 2 resolution methods (based on client resolver)
- Recursive resolution: ISP DNS resolver will resolve the name
- Root DNS server: direct to the Top-Level Domain (TLD) server (e.g. .com, .edu, .org, .sg)
- TLD server: direct to the Authoritative DNS Server for the domain
- Authoritative DNS Server: returns actual IP address
- Iterative resolution: client-resolver / local ISP DNS resolver performs queries step by step, using referrals from each server
- Root DNS server: returns address of a TLD server
- TLD server: returns address of authoritative DNS server
- Authoritative DNS Server: returns actual IP address
- Return IP address to client
- DNS resolver sends the resolved IP address back to client
- Client uses IP address to route the message in the network
- Advantages of DNS using UDP
- Speed
- Fast as users expect pages to load fast
- UDP does not need to establish a connection setup
- Small size
- DNS queries and responses easily fit in 1 UDP datagram
- Low overhead
- Simple header
- Requires less processing time and bandwidth
- Stateless
- Simple request and response
- Does not need to maintain connection
Hypertext Transfer Protocol (HTTP)
- Note: using HTTP != HTML (i.e. it must not be for a webpage
- Allows clients (e.g. browsers) to fetch hypertext documents, images, scripts, etc from web server
- Uses ports 80, 443 (HTTPS)
- Uses client-server model
- Client sends a HTTP request
- Server responds with a HTTP response
- ⇒ these 2 separate requests are treated independently
- A stateless protocol ⇒ server does not keep information about client
- Each request-response pair is independent
- Client initializes the transaction by sending a request
- Server replies by sending a response
- HTTP transaction between client and server
- HTTP request includes:
- Request line:
GET /test.html HTTP/1.1
xGET: HTTP command/method that the client wants to performtest.html:path of the relevant documentHTTP/1.1: version of the HTTP used by the client- Header fields: provides additional info to web server
- E.g.
Host: example.com - Message body (optional): for POST requests, message body contains the data submitted
- HTTP response includes:
- Status line: summarises whether server is able to perform request from client
- E.g.
HTTP/1.1 200 OK HTTP/1.1: HTTP version used by server- Status code: reason phrase
- 200: OK; 404: Not found; 500: Internal Server Error
- Header fields: provides additional info
- E.g.
Content-Type: text/plain - Message body: contains document in format specified by
Content-Typeheader field 
Simple Mail Transfer Protocol (SMTP)
- Protocol used to send/relay email across the internet, not for rerieving email
- Uses Port 25/587 (secure)
- Uses client-server model
- SMTP Message Structure:
- Envelope (Control Information)
- MAIL FROM: sender address
- RCPT TO: recipient address
- DATA: begin message content
- Message Headers: describes sender, receiver, content type, etc
- Message Body: content of email
Post Office Protocol v3 (POP3)
- Protocol to download emails from server to client and delete the message from the mailbox on the server
- Uses port 110/995 (secure)
- Uses client-server model: email client and server
 | Mail access starts with the client Client opens a connection to the server on TCP port 110 It then sends its user name and password to access the mailbox The user can then list and retrieve the mail messages, 1 by 1 |
|---|
Internet Message Access Protocol v4
- Similar to POP3 and allows user to view and manage messages on server
- Allows user to access mail messages from multiple locations (email clients) and ensure that all copies are synchronized and consistent
- Extended functionality:
- Obtain info about the message or examine header fields without retrieving the entire message
- Search and retrieve portions of a message (partial retrieval)
- Useful for slow-speed connections
- Uses Port 143/993 (secure)
- Microsoft Outlook, Apple Mail, Mozilla Thunderbird, Webmail email client uses IMAPv4
File Transfer Protocol (FTP)
- Protocol for copying a file from one host to another
- Resolves some problems with 2 hosts having:
- Diff. file name conventions
- Diff. ways of representing text and data
- Diff. directory structures
- Differs from other client-server applications as it establishes 2 connections between the hosts
- Control Connection / information (port 21)
- Used for commands from clients and response from server
- Remains open during the session
- Data Connection / transfer (port 20)
- Used for file transfer: upload, download, and directory listing
- Open when needed, and closed after data transfer
