Routing and Switching: Hub, Switch, Router, VLANs and Routing Protocols · রাউটিং ও সুইচিং: হাব, সুইচ, রাউটার, ভিল্যান ও রাউটিং প্রোটোকল

1. The devices: repeater, hub, bridge, switch, router, gateway

Each interconnecting device works at a particular OSI layer, and that layer decides what it can "see" (the layers are explained on the OSI and TCP/IP models page).

Device	Layer	Forwards using	What it does	Collision domains	Broadcast domains
Repeater	1 (Physical)	Nothing (just signals)	Regenerates a weakened signal to extend a cable's reach	Does not split	Does not split
Hub	1	Nothing	A multiport repeater: copies every bit out of every other port	1 for the whole hub	1
Bridge	2 (Data Link)	MAC address	Joins two (or a few) segments and filters frames between them; the segments still look like one network to higher layers	1 per port	1
Switch	2	MAC address	A multiport bridge with hardware forwarding; sends each frame only to the port where the destination is	1 per port	1 (unless VLANs are used)
Router	3 (Network)	IP address	Joins different networks and chooses the path for each packet	1 per interface	1 per interface
Gateway	Up to 7	Depends on the protocols	Connects networks that use different protocols and translates between them	–	–

Points that exams like:

• A switch has memory: it stores a MAC address table and buffers frames. "A switch has no memory" is the wrong statement in MCQs.
• A managed switch can be configured port by port, for example to create VLANs; an unmanaged switch cannot.
• A layer-3 switch is a switch that can also route between VLANs.
• Default gateway has a narrower meaning: the router a host sends packets to when the destination is outside its own subnet. In everyday use "gateway" and "router" overlap, but a true gateway also converts between protocols (for example email or VoIP gateways).
• A firewall filters traffic by security rules; a gateway connects and translates. A modem converts digital data to signals for a telephone, cable or fibre line, while a router chooses paths between IP networks (home "routers" usually combine a modem, router, switch and Wi-Fi access point in one box).

2. Collision domains and broadcast domains

• A collision domain is the set of devices whose transmissions can collide because they share one medium (half duplex). Only one device in it can send at a time.
• A broadcast domain is the set of devices that receive each other's broadcast frames (such as ARP requests or DHCP Discover).

Rule: hubs split nothing, switches and bridges split collision domains, routers split collision domains and broadcast domains. Each switch port is its own collision domain, so a 12-port switch gives 12 collision domains but only 1 broadcast domain.

Worked example 1: counting domains. A router has three interfaces:

• interface 1 connects to a 24-port switch; 8 of its ports are in use: one uplink to the router, six PCs, and one hub that has 4 PCs on it;
• interface 2 connects to a hub with 5 PCs;
• interface 3 is a serial link to an ISP router.

1. Broadcast domains: count router interfaces: 3.
2. Collision domains on the switch: every used switch port is one: the uplink to the router (1), six PCs (6), and the hub port (1, which includes the hub and all 4 of its PCs, because a hub does not split collisions). That is 8.
3. Interface 2: the hub, its 5 PCs and the router interface share one collision domain: 1.
4. Interface 3: the serial link is 1.
5. Total collision domains = 8 + 1 + 1 = 10.

The shortcut for the simple MCQ form: N switch ports → N collision domains, 1 broadcast domain; N hub ports → 1 and 1; N router interfaces → N and N.

3. How a switch works

A switch keeps a MAC address table (often called the CAM table) mapping MAC addresses to ports, and builds it by itself:

1. Learn: for every incoming frame it records source MAC → incoming port.
2. Forward or filter: if the destination MAC is in the table, it sends the frame only out of that port (and drops it if that port is the one it came in on).
3. Flood: if the destination is unknown, or is the broadcast address FF-FF-FF-FF-FF-FF, it sends the frame out of every port except the incoming one.
4. Age: entries not seen for a while (commonly 300 seconds) are removed, so the table follows devices that move.

Worked example 2: the table filling up. PCs A, B and C are on ports 1, 2 and 3 of a new switch with an empty table.

Frame	Switch learns	Destination known?	Action
A → B	A is on port 1	No	Flood to ports 2 and 3
B → A	B is on port 2	Yes (port 1)	Forward only to port 1
C → broadcast	C is on port 3	Broadcast	Flood to ports 1 and 2
A → C	(A already known)	Yes (port 3)	Forward only to port 3

Switching methods.

Method	Starts forwarding after	Error checking	Delay
Store-and-forward	The whole frame is received and its CRC (FCS) checked	Drops damaged frames	Highest: a 1,518-byte frame at 100 Mbps adds 121.44 µs
Cut-through	The destination MAC address (first 6 bytes) is read	None: bad frames are passed on	Lowest
Fragment-free	The first 64 bytes are received (where collisions show up)	Catches collision fragments only	In between

MAC flooding is an attack in which a host sends thousands of frames with fake source MACs until the table is full; the switch then floods unknown frames like a hub and the attacker can capture other people's traffic. It is prevented with port security (limit the number of MAC addresses per port and shut down or block violators), 802.1X port authentication, and by disabling unused ports.

4. VLANs and Spanning Tree

A VLAN (virtual LAN) divides one physical switch network into several logical networks, each its own broadcast domain. Departments such as Accounts and HR can share switches yet stay separated, which cuts broadcast traffic, improves security and makes moves and changes easier. VLANs reduce management complexity, so "VLANs increase network management complexity" is the false statement in MCQs. VLANs are created on switches, and traffic between VLANs must pass through a router or layer-3 switch.

• IEEE 802.1Q adds a 4-byte tag to Ethernet frames on trunk links (links carrying many VLANs); its 12-bit VLAN ID gives 4,096 values, of which 1 – 4,094 are usable (0 and 4,095 are reserved). An access port belongs to one VLAN and sends untagged frames.
• Static (port-based) VLAN: the administrator assigns each port to a VLAN; simple and the most common. Dynamic VLAN: a device is placed in a VLAN automatically by its MAC address or login, using a policy server, so it keeps its VLAN wherever it plugs in.

Spanning Tree Protocol (STP, IEEE 802.1D, now part of 802.1Q). Redundant links between switches create loops, and since Ethernet frames have no TTL, a broadcast would circle forever (a broadcast storm) and MAC tables would keep changing. STP prevents this:

1. The switches elect a root bridge: the one with the lowest bridge ID (priority, default 32,768, then MAC address).
2. Every other switch picks its root port, the port with the lowest path cost to the root.
3. Each segment gets one designated port; all remaining ports are put into blocking state, leaving a loop-free tree.
4. If an active link fails, a blocked port moves through listening and learning to forwarding. With the default timers (hello 2 s, max age 20 s, forward delay 15 s) this takes 30 to 50 seconds; Rapid STP (802.1w) usually converges in a few seconds.

5. Routing: the routing table and longest-prefix match

A router reads a packet's destination IP address, looks it up in its routing table, and sends the packet to the next hop through an outgoing interface. It also decrements the TTL and builds a new layer-2 frame, so the IP addresses stay the same end to end (unless NAT rewrites them) while the MAC addresses change at every hop; a MAC address never leaves the local network. To build that frame the router needs the next hop's MAC address, which it finds with ARP (and keeps in its ARP cache), just as a host does for its default gateway.

Each routing table entry holds a destination network and prefix, a next hop or exit interface, and a metric. When several entries match, the router uses the most specific one: the entry with the longest prefix (longest-prefix match). The default route 0.0.0.0/0 matches every address with a prefix length of 0, so it is used only when nothing else matches (the "gateway of last resort"). The address arithmetic is explained on the IP addressing and subnetting page.

Worked example 3: longest-prefix match.

Destination	Next hop
10.0.0.0/8	A
10.1.0.0/16	B
10.1.4.0/22	C
10.1.5.128/25	D
0.0.0.0/0	E

1. 10.1.5.200: /8 matches (10.x.x.x), /16 matches (10.1.x.x), /22 matches (10.1.4.0 – 10.1.7.255), /25 matches (10.1.5.128 – 10.1.5.255), and /0 always matches. The longest is /25 → D.
2. 10.1.5.20: inside 10.1.4.0/22 but below .128, so not in the /25 → C.
3. 10.1.6.9: the /22 covers 10.1.4.0 – 10.1.7.255 → C.
4. 10.2.3.4: only /8 and /0 match → A.
5. 11.0.0.1: only the default route matches → E.

6. Static vs dynamic routing

Point	Static routing	Dynamic routing
How routes appear	Typed in by an administrator	Learned from other routers by a routing protocol
Reacts to failures	No, until someone edits it	Yes, automatically
Bandwidth and CPU use	None (no updates sent)	Uses some for updates and calculations
Security	Higher (nothing advertised)	Lower unless updates are authenticated
Best for	Small or stub networks, default routes to an ISP	Medium and large networks with redundant paths

A typical static default route on a Cisco router toward an ISP: ip route 0.0.0.0 0.0.0.0 203.0.113.1 (destination, mask, next hop). "Which routing does not consume network bandwidth?" is keyed static routing. Routing in which each router decides by itself on the fly, as in mobile ad hoc networks with no fixed infrastructure, uses on-demand protocols such as AODV that find routes only when needed, saving bandwidth and battery.

7. Distance vector, link state and path vector

An autonomous system (AS) is a group of networks and routers under one administration (an ISP, a large organisation). Interior gateway protocols (IGPs) route inside an AS (intra-domain): RIP, OSPF, IS-IS, EIGRP. Exterior gateway protocols route between ASes (inter-domain): BGP.

Point	Distance vector (RIP)	Link state (OSPF)	Path vector (BGP)
What a router knows	Only distances its neighbours report ("routing by rumour")	A full map of the area	Full AS paths to each network
What it sends	Its whole table to its neighbours, periodically (RIP: every 30 s)	Link-state advertisements flooded to all routers in the area when something changes (and refreshed every 30 minutes)	Updates to its peers over TCP 179
Algorithm	Bellman-Ford	Dijkstra's shortest path first (SPF)	Best-path selection by policy and AS-path
Metric	Hop count (maximum 15; 16 = unreachable)	Cost, based on bandwidth	Path attributes (AS-path length, policies)
Convergence	Slow; can count to infinity	Fast	Slow, but stable at Internet scale
Scope	Small networks (IGP)	Large networks with areas; area 0 is the backbone (IGP)	Between autonomous systems: the Internet (EGP)

Cisco's EIGRP is an advanced distance-vector protocol (sometimes called "hybrid") using the DUAL algorithm; IS-IS is another link-state IGP.

Distance vector loops and their fixes. When a link fails, neighbours can keep advertising the dead route to each other, increasing it step by step (count to infinity). Fixes: a maximum hop count (RIP's 16 = infinity), split horizon (do not advertise a route back out of the interface it was learned on), route poisoning (advertise a failed route with an infinite metric, 16 in RIP), poison reverse (advertise it back with infinity), and hold-down timers.

RIPv1 vs RIPv2. RIPv1 is classful (its updates carry no subnet mask) and broadcasts them; RIPv2 is classless (it carries the mask, so it supports VLSM, and adds authentication) and multicasts its updates to 224.0.0.9. OSPF details: OSPF runs directly over IP (protocol number 89), sends Hello packets to 224.0.0.5 to find and keep neighbours, and on a multi-access network such as Ethernet elects a designated router (DR) and a backup (BDR), reached at 224.0.0.6, so that every router exchanges LSAs with the DR instead of with every other router.

Worked example 4: a distance-vector update. Router A has links to B (cost 2) and C (cost 5). B reports: C at 1, D at 3. C reports: B at 1, D at 1. A takes, for each destination, the minimum of link cost + neighbour's distance:

• to D: via B = 2 + 3 = 5, via C = 5 + 1 = 6 → 5 via B;
• to C: via B = 2 + 1 = 3, direct = 5 → 3 via B (the direct link is not the best path);
• to B: direct 2, via C = 5 + 1 = 6 → 2 direct.

Worked example 5: Dijkstra (as OSPF uses it). Links and costs: A–B 4, A–C 2, B–C 1, B–D 5, C–D 8, C–E 10, D–E 2. From A:

1. Start: A = 0; all others unknown (∞).
2. Fix A. Neighbours: C = 2, B = 4.
3. Fix the smallest, C (2). Through C: B = 2 + 1 = 3 (better than 4), D = 2 + 8 = 10, E = 2 + 10 = 12.
4. Fix B (3). Through B: D = 3 + 5 = 8 (better than 10).
5. Fix D (8). Through D: E = 8 + 2 = 10 (better than 12).
6. Fix E (10). Done.

Shortest paths: C = 2 (A-C), B = 3 (A-C-B), D = 8 (A-C-B-D), E = 10 (A-C-B-D-E).

OSPF cost. OSPF's metric is cost. On Cisco routers cost = 10⁸ ÷ interface bandwidth in bps (rounded down, minimum 1): 10 Mbps → 10, a 1.544 Mbps T1 → 64, 100 Mbps and faster → 1 (which is why faster links need a larger reference bandwidth).

8. Administrative distance and tricky exam cases

When two different sources (say OSPF and RIP) offer a route to the same network, the router trusts the one with the lower administrative distance (AD). AD values are Cisco defaults, not an Internet standard:

Route source	Default AD
Directly connected	0
Static route	1
External BGP	20
Internal EIGRP	90
OSPF	110
IS-IS	115
RIP	120
External EIGRP	170
Internal BGP	200
Unknown (never used)	255

Order of decisions: longest prefix first, then lowest AD between protocols, then lowest metric within one protocol. A /24 learned by RIP beats a /16 learned by OSPF, because prefix length is compared before AD.

• Hub bandwidth. Ten users on a 10 Mbps half-duplex hub share 10 Mbps, so each gets about 10 ÷ 10 = 1 Mbps. On a switch, each port would get the full 10 Mbps.
• "How many pairs of stations can communicate at once on an Ethernet LAN?" Keyed 1: classic shared Ethernet (a bus or hub) is one collision domain. With a full-duplex switch, many pairs can talk at once; read the question.
• "When a host on network A sends to network B, which address does the router look at?" The destination IP address. A MAC address stays on the local LAN.
• Pairs of routing protocols. Intra-domain: RIP and OSPF. Routing protocols in general: RIP, OSPF, EIGRP, BGP. Routed protocol: IP. BGP uses the AS-path to build its table. RIP uses Bellman-Ford; OSPF uses Dijkstra.
• Ethernet is IEEE 802.3; Wi-Fi is 802.11; VLAN tagging is 802.1Q.