PROJECTENGRAM

APPLICATIONS

Applications

Tags: #◇Computer Science #◇CSNETWK #◇Applications #◇Code Snippets

◇Overview

The Application Layer is the top layer where network applications and their application-layer protocols reside. The development of network applications focuses on end systems without writing code for the network core, taking advantage of rapid application development.

Network Paradigms Client-Server Architecture: A server is an always-on host with a permanent IP address that waits to be contacted, often scaled via data centers. Clients intermittently connect to the server and do not directly communicate with each other (e.g., HTTP, IMAP, FTP). Peer-to-Peer (P2P) Architecture: There is no always-on server. Arbitrary end systems directly communicate, bringing self-scalability since each peer adds service capacity and demand. Examples include BitTorrent.

Inter-process Communication and Sockets

Processes (programs running on a host) communicate across different hosts via messages. They send and receive messages through sockets—the interface (or "door") between the application layer and transport layer. Processes are identified using a combination of the host's 32-bit IP address and a port number (e.g., Port 80 for an HTTP server, Port 25 for a mail server).

Application-Layer Protocols

Protocols define the types of messages exchanged (request/response), syntax (fields and their delimiters), semantics (meaning of the fields), and rules for how processes respond. Open protocols: Defined in RFCs for interoperability (HTTP, SMTP). Proprietary protocols: Closed definitions (Skype, Zoom).

Transport Services Required by Applications

Applications vary in their transport requirements: Data Integrity: Some require 100% reliable transfer (file transfers, web), others are loss-tolerant (audio). Timing: Some require low delay (Internet telephony, interactive games). Throughput: Some need a minimum bandwidth (multimedia), others are "elastic" and use what is available. Security: Needs for encryption and data integrity.

TCP provides reliable, in-order delivery, flow control, and congestion control. UDP provides unreliable, best-effort datagram delivery with no connection setup.

The Web and HTTP HTTP (Hypertext Transfer Protocol): The Web's application-layer protocol, based on a client/server model. It is a stateless protocol meaning it maintains no information about past client requests. It typically uses TCP on port 80. HTTP Connections:

Non-persistent: Opens a new TCP connection for each object.
Persistent: Multiple objects sent over a single TCP connection. HTTP Requests: Includes methods like GET (requests object), POST (sends user input), HEAD (requests headers only), PUT (uploads object). HTTP Responses: Includes status codes (e.g., 200 OK, 301 Moved Permanently, 400 Bad Request, 404 Not Found, 505 Version Not Supported). Cookies: Used by sites to maintain state. Components include cookie headers in HTTP messages, a local cookie file on the client, and a back-end database on the server. Web Caches (Proxy Servers): Intermediaries that store recently requested objects to satisfy client requests locally. This reduces response time and traffic on the institution's access link. A Conditional GET (using If-modified-since) helps keep caches up to date. HTTP/2 & HTTP/3: HTTP/2 mitigates head-of-line (HOL) blocking by dividing objects into frames and interleaving them. HTTP/3 moves transport to UDP to prevent TCP loss recovery from stalling all object transmissions and adds integrated security.

E-mail and DNS E-mail components: User agents (readers), mail servers (mailbox and outgoing queue), and SMTP (Simple Mail Transfer Protocol, used between mail servers to send messages). DNS (Domain Name System): A distributed, hierarchical database mapping human-readable hostnames to IP addresses. It consists of root DNS servers, Top-Level Domain (TLD) servers (like .com, .edu), and authoritative servers. Queries can be iterated or recursive. DNS aggressively uses caching (with a TTL) to improve response times.

◇Important Formulas

(No specific formulas were discussed in this module)

◇Worked Examples

Example 1: Web Cache Delay Analysis
Given an access link rate of 1.54 Mbps, average object size of 100K bits, and a request rate of 15 requests/sec.
Without cache:
Data rate to browsers = 15×100 Kbits/sec=1.5 Mbps15 \times 100 \text{ Kbits/sec} = 1.5 \text{ Mbps}.
Access link utilization = 1.5/1.54=0.971.5 / 1.54 = 0.97 (97%), leading to high queueing delays (minutes).
With cache (40% hit rate):
The request rate going to the origin server drops to 60%.
Data rate on access link = 0.6×1.5 Mbps=0.9 Mbps0.6 \times 1.5 \text{ Mbps} = 0.9 \text{ Mbps}.
Access link utilization = 0.9/1.54=0.580.9 / 1.54 = 0.58 (58%). Delay drops to milliseconds.
Average end-to-end delay = 0.6×(origin delay)+0.4×(cache delay)0.6 \times (\text{origin delay}) + 0.4 \times (\text{cache delay}).

◇References & Navigation

CONFIRMATION // 理解度の確認

ACTIVE RECALL VERIFICATION

Select an answer to verify conceptual comprehension and reveal the technical rationale.

Q1

In the Client-Server vs. Peer-to-Peer (P2P) application architectures, how does the minimum distribution time to deliver a file of size FF to NN peers scale as NN grows large?

Q2

How many Round-Trip Times (RTTs) are required to fetch a base HTML page plus 1010 referenced JPEG objects over Non-Persistent HTTP/1.0 (without parallel connections) versus Persistent HTTP/1.1 with pipelining?

Q3

In the Domain Name System (DNS) hierarchy, what is the difference between a recursive query and an iterative query?

Q4

Which DNS Resource Record (RR) type maps a hostname alias (e.g., www.example.com) to its canonical hostname, and which RR type maps a domain to the hostname of its mail server?

Q5

Why does FTP (File Transfer Protocol) described as maintaining an out-of-band control connection, whereas HTTP sends control information in-band?