A computer network is a system where two or more computing devices are connected to share resources, exchange data, and communicate. Think of it like the Indian postal system — you need addresses (IP addresses), post offices that sort and route mail (routers and switches), agreed-upon procedures for packaging letters (protocols), and different categories of routes — local courier (LAN), citywide delivery (MAN), national courier (WAN).
Networks exist at different scales:
The key concepts IBPS PO tests repeatedly are: OSI model layer assignments, which device operates at which layer, protocol-to-function mapping (what sends email, what secures file transfer), and basic subnetting arithmetic. These are factual recall questions — you either know the mapping or you don't. This page gives you the exact mappings and a framework to retain them.
The two dominant network architecture models are OSI (Open Systems Interconnection) and TCP/IP. OSI is the theoretical reference model with 7 layers. TCP/IP is the practical model the internet actually runs on, with 4 layers. IBPS PO questions almost always use OSI layer numbers and names — nail those.
The OSI model layers, from bottom to top:
| Layer | Number | Name | Key Function | |---|---|---|---| | Physical | 1 | Physical | Bits over wire/fiber/radio | | Data Link | 2 | Data Link | Frames, MAC addresses, error detection | | Network | 3 | Network | Packets, IP addresses, routing | | Transport | 4 | Transport | Segments, TCP/UDP, end-to-end delivery | | Session | 5 | Session | Session management, dialog control | | Presentation | 6 | Presentation | Encryption, compression, format translation | | Application | 7 | Application | User-facing protocols: HTTP, FTP, SMTP |
The mnemonic "Please Do Not Throw Sausage Pizza Away" (Physical, Data Link, Network, Transport, Session, Presentation, Application) works fine. But more importantly — understand why each device maps where it does.
This is the single most-tested pairing in IBPS PO computer sections:
The rule: higher layer = more intelligence = knows more about addressing. Hub knows nothing, switch knows MAC, router knows IP.
Learn these as hard fact-pairs:
Application Layer Protocols (Layer 7):
The email protocol distinction matters: SMTP is for sending. POP3 and IMAP are for receiving/reading. The exam loves asking which is for sending vs. receiving.
Transport Layer Protocols (Layer 4):
IPv4 addresses are 32-bit numbers written as four octets: 192.168.1.1. CIDR notation adds a slash and the number of network bits: 192.168.1.0/24 means the first 24 bits are the network portion.
Host calculation formula: 2^(32 - prefix) - 2
The -2 accounts for the network address (all host bits = 0) and broadcast address (all host bits = 1) — neither can be assigned to a host.
For /24: 2^(32-24) - 2 = 2^8 - 2 = 256 - 2 = 254 usable hosts.
IP Address Classes (default subnet masks):
| Class | Range (1st octet) | Default Mask | CIDR | |---|---|---|---| | A | 1–126 | 255.0.0.0 | /8 | | B | 128–191 | 255.255.0.0 | /16 | | C | 192–223 | 255.255.255.0 | /24 |
Class A: huge networks, few network IDs, many hosts. Class C: small networks, many network IDs, few hosts. Class B sits in the middle — common in mid-size organizations like bank networks.
Lock in three device-layer pairs as a chain: Hub=1, Switch=2, Router=3. That's it — Physical, Data Link, Network respectively. In the exam, any question asking "which device operates at which layer" can be answered in under 5 seconds with this triple. Every other device (bridge=2, repeater=1, gateway=7) is a variant you derive from this anchor. Standard recall without anchor: 20-25 seconds of uncertainty. With the 1-2-3 lock: under 5 seconds.
Remember: SMTP = Sending. POP3 and IMAP both = receiving (and both come after SMTP alphabetically and operationally). When you see an email protocol question, first ask "is this about sending or receiving?" — if sending, SMTP is the only answer. If the question distinguishes POP3 from IMAP: POP3 Pulls mail and Purges it from server; IMAP keeps it on server. This pattern eliminates wrong options in 8 seconds vs. 30 seconds of trying to recall definitions cold.
Given /X notation — Step 1: host bits = 32 - X. Step 2: total addresses = 2^(host bits). Step 3: subtract 2. That's your answer. For /24: host bits = 8, total = 256, usable = 254. For /16: host bits = 16, total = 65536, usable = 65534. This 3-step chain takes 10 seconds. The common mistake is forgetting to subtract 2 — always do it. Without the chain, students second-guess whether to subtract 1 or 2 and lose 30+ seconds.
Any protocol with an S prefix (SFTP, HTTPS, SSH) means the secure version of the base protocol. SFTP = FTP + encryption. HTTPS = HTTP + SSL/TLS. SSH = Telnet + encryption. When a question asks "which protocol provides secure file transfer," the answer is always the S-prefixed version. This pattern resolves the question in under 5 seconds — you don't need to recall port numbers or technical details.
First octet 1-126 = Class A. First octet 128-191 = Class B. First octet 192-223 = Class C. Quick anchor: 127 is loopback (not Class A usable). 128 starts Class B — think of it as "B for Big jump" at 128. 192 starts Class C — think "C for Common office network" (192.168.x.x is everywhere). Default mask question: Class A gets /8, Class B gets /16, Class C gets /24 — the class number times 8 gives the CIDR prefix. Class B: 2×8 = /16. Class C: 3×8 = /24. This cuts mask lookup to a multiplication.
When you see a computer networking question in the exam hall, run this decision tree:
Step 1 — Is it a device/layer question? Apply the 1-2-3 lock immediately. Hub=1, Switch=2, Router=3. Done.
Step 2 — Is it a protocol question? Ask: sending email (SMTP), receiving email (POP3/IMAP), secure file transfer (SFTP), web (HTTP/HTTPS), secure remote login (SSH). Check if the question asks about security — if yes, eliminate unencrypted options (FTP, Telnet, HTTP) first.
Step 3 — Is it a subnetting question? Apply the 3-step chain: host bits = 32 minus prefix, total = 2 to the power of host bits, subtract 2. If it's a Class question, use the class times 8 rule for the CIDR prefix.
Step 4 — Is it a topology or network type question? Star = most common modern LAN. Mesh = maximum redundancy. LAN/MAN/WAN = building/city/national scope. These are pure recall.
Step 5 — Eliminate first, then confirm. On 4-option MCQs, if you can eliminate two options using these rules, your odds go from 25% to 50% before you use any deeper knowledge. Never leave a networking question blank — elimination alone gets you far.
Why this question: Tests the single most common subnetting calculation in IBPS PO — host count from CIDR notation. Appears in bank exams repeatedly because it has one clean correct answer with a common trap (256 instead of 254).
Solving path: Apply 3-step chain. Prefix = 24. Host bits = 32 - 24 = 8. Total addresses = 2^8 = 256. Subtract 2 (network + broadcast) = 254. The trap option is 256 — correct only if you forget to subtract 2. The trap option 255 has no mathematical basis. Answer: 254.
Why this question: Tests a pure-recall hardware specification. Ethernet cable distance limits come up in IBPS PO as a quick elimination question — most candidates don't know the exact figure.
Solving path: The 100-meter limit for Cat5e/Cat6 is a standard specification to memorize. The logic: signal degrades with distance; 100m is the engineering standard beyond which you need a repeater or switch. Eliminate 50m (too short — practical office networks often exceed this), 150m and 200m (both exceed the standard). Answer: 100 meters.
Why this question: Tests protocol-to-function mapping for network diagnostics. Ping, tracert, nslookup, and netstat are all network commands but serve completely different diagnostic purposes.
Solving path: ping sends ICMP echo requests to test basic reachability. tracert (traceroute on Linux) shows the path packets take — hop by hop. nslookup resolves domain names to IP addresses. netstat shows active connections and listening ports. The question asks specifically about testing connectivity between two hosts — that's ping's primary job. Answer: ping.
Why this question: Tests the email protocol direction distinction — one of the highest-frequency protocol questions in IBPS PO computer sections.
Solving path: Eliminate POP3 and IMAP immediately — both are for retrieving/reading mail, not sending. Eliminate HTTP — it's a web protocol, unrelated to email. SMTP is specifically the protocol for outbound email delivery: client to mail server, and between mail servers. Answer: SMTP.
Why this question: Tests OSI layer-to-device mapping — one of the top two most-tested networking concepts in IBPS PO computer awareness sections.
Solving path: Apply the 1-2-3 lock. Hub = Layer 1. Switch = Layer 2. Router = Layer 3. The question asks for the Data Link Layer (Layer 2) device. Switch is the answer. Gateway operates at higher layers (often Layer 7). Router is Layer 3, not Layer 2. Answer: Switch.
Confusing 256 with 254 in subnetting. The /24 network has 256 total addresses but only 254 usable — students forget to subtract the network and broadcast addresses. Always subtract 2 after the 2^n calculation.
Mixing up SMTP, POP3, and IMAP. SMTP sends mail. POP3 and IMAP receive it. A common wrong answer is selecting POP3 for "sending email" because POP3 is the best-known email protocol name. Remember: S in SMTP = Sending.
Placing Router at Layer 2 instead of Layer 3. Routers use IP addresses (Layer 3). Switches use MAC addresses (Layer 2). The distinction matters because MAC addresses are local to a single network; IP addresses span multiple networks — which is exactly what routers do.
Selecting FTP instead of SFTP for secure file transfer. FTP transmits data in plain text — it has no encryption. Any exam question that includes the word "secure" in "file transfer" points to SFTP. FTP is never the correct answer when security is specified.
Confusing Class B default mask with Class C. Class B = 255.255.0.0 (/16). Class C = 255.255.255.0 (/24). The mistake usually goes the other direction — assigning /24 to Class B. Use the class-times-8 rule to avoid this.
Treating Hub and Switch as interchangeable. They are not. Hub broadcasts to all ports (Layer 1, one collision domain). Switch forwards only to the correct port using MAC tables (Layer 2, separate collision domain per port). Exam questions that ask about "creating separate collision domains" are testing whether you know the Switch, not the Hub.