Network Layer Protocol Practice Questions

CIDR block 201.70.64.0/18: host bits = 32 - 18 = 14 bits. Total IP addresses = 2^14 = 16,384. This includes 1 network address (all host bits = 0) and 1 broadcast address (all host bits = 1). Usable host addresses = 16,384 - 2 = 16,382. The block spans from 201.70.64.0 to 201.70.127.255 (64 Class C networks, each with 256 addresses: 64 × 256 = 16,384).

The count-to-infinity problem occurs in Distance Vector Routing when a link failure causes routers to keep incrementing the distance to the failed destination, slowly counting up to infinity (defined as 16 in RIP).

RFC 1918 defines three private IP address ranges not routable on the public internet: (1) Class A private range: 10.0.0.0 to 10.255.255.255 (10.0.0.0/8) - 16,777,216 addresses. (2) Class B private range: 172.16.0.0 to 172.31.255.255 (172.16.0.0/12) - 1,048,576 addresses. (3) Class C private range: 192.168.0.0 to 192.168.255.255 (192.168.0.0/16) - 65,536 addresses. These ranges are used for private networks (homes, offices) and require NAT to communicate with the public internet. Special addresses: 127.0.0.0/8 (loopback), 169.254.0.0/16 (link-local/APIPA).

OSPF hierarchical routing with areas: A: TRUE - Area 0 is the backbone; all other areas (Area 1, 2, ...) must have at least one connection to Area 0. This star topology of areas ensures all inter-area traffic passes through the backbone. B: TRUE - LSA flooding is contained within a single area, preventing the entire network from being flooded. ABRs summarize inter-area routing information (Type 3 LSAs), drastically reducing LSDB size in large networks. C: TRUE - ABRs (Area Border Routers) have interfaces in multiple areas; they receive detailed LSAs from each area and generate summary LSAs for other areas. D: FALSE - OSPF always runs Dijkstra's algorithm within each area; areas reduce the size of the topology graph on which SPF runs, not eliminate it.

Link State routing uses Dijkstra.

Hierarchical routing: routers are grouped into regions/areas/ASes. Within a region, routers know the full topology. Between regions, only summary information is exchanged. Advantages: (1) Reduced routing table size - routers only store detailed routes for their region and summary routes for other regions. (2) Reduced LSA flooding - link-state updates don't propagate beyond area boundaries (in OSPF). (3) Scalability - the internet cannot run flat link-state routing (billions of prefixes would require exabytes of memory). Disadvantages: (1) Routes may be non-optimal - summarization loses detailed topology information, so the chosen path may not be the globally shortest. (2) Configuration complexity - area design, ABR configuration, summarization policies. D is FALSE - hierarchical routing reduces memory usage.

Link-state routing protocol steps (OSPF): (1) Neighbor discovery - routers send Hello packets on each interface; routers that exchange Hellos become neighbors. (2) LSA creation - each router generates a Link State Advertisement describing its router ID, its neighbors, and the cost to each neighbor. (3) LSA flooding - each router reliably floods its LSA to every other router in the OSPF area; each router forwards each LSA exactly once (using sequence numbers to detect duplicates). (4) Link State Database (LSDB) - each router builds an identical topology graph from the collected LSAs. (5) SPF calculation - each router independently runs Dijkstra's Shortest Path First (SPF) algorithm on the LSDB, computing the shortest-path tree rooted at itself. (6) Routing table - the SPF tree's next-hops populate the routing table.

Bellman-Ford finds shortest paths in at most V-1 = 4 iterations for V = 5 vertices. Let's trace: the shortest path A→B→C→D→E has cost 1+2+1+1=5 (better than direct A-E=10). Iteration 1: d[B]=1(A-B), d[E]=10(A-E). Iteration 2: d[C]=3(A-B-C), d[E]=10 (no improvement via B-E). Wait, there is no B-E link. Iteration 2: d[C]=1+2=3. Iteration 3: d[D]=3+1=4. Iteration 4: d[E]=min(10, 4+1)=5. The shortest path A-B-C-D-E=5 < A-E=10 is found at iteration 4. Bellman-Ford requires 4 = V-1 iterations because the shortest path A→E has 4 edges.

SDN architecture vs traditional routing: Traditional: each router has both a control plane (routing software running protocols like OSPF, BGP) and a data plane (hardware forwarding using routing table). Distributed computation - each router independently runs routing algorithms. SDN: control plane is separated from the data plane. Centralized SDN controller (logically centralized, physically distributed) has a global network view and computes routing/forwarding rules for ALL switches. Switches (data plane) only have flow tables installed by the controller via a southbound API (e.g., OpenFlow) - they perform simple match-action forwarding. Benefits: global optimization, simpler network devices, easier policy enforcement, programmability. Examples: Google B4 (SDN WAN), data center fabrics.

Subnet mask /26 = 255.255.255.192. Binary of last octet: 130 = 10000010. Mask last octet: 192 = 11000000. Network address (AND operation): 130 AND 192 = 10000010 AND 11000000 = 10000000 = 128. So network address = 172.16.45.128. Broadcast address: set all host bits to 1. Host bits = 6 bits (32-26=6). Network last octet = 128 = 10000000. Broadcast = 10000000 OR 00111111 = 10111111 = 191. Broadcast = 172.16.45.191. Usable hosts: 172.16.45.129 to 172.16.45.190 (62 hosts).

BGP policy-based routing examples: A: TRUE - ISPs establish business relationships (customer, peer, provider). Routing policies enforce: traffic from customers can be forwarded to providers and peers (customer pays); peer traffic only forwarded to/from own customers (settlement-free peering). A router preferring a business partner's path over a competitor's uses LOCAL_PREF or WEIGHT attributes. B: TRUE - LOCAL_PREF is set by the BGP router and propagated to all iBGP peers within the same AS. Higher LOCAL_PREF = preferred exit. Used to implement 'primary/backup' exit policies. C: TRUE - MED (Multi-Exit Discriminator) is sent to neighboring ASes to suggest which entry point to use when they have multiple connections to your AS. Lower MED = preferred entry. D: FALSE - BGP's primary purpose IS policy-based routing; it explicitly allows (and requires) policies for proper operation.

Split horizon rule: if router A learned that destination D is reachable via router B (next hop = B), then A will NOT include destination D in its routing updates sent back to B. Rationale: B already knows the best path to D (it told A about it), so advertising it back is useless and can cause count-to-infinity. Simple split horizon (option A) suppresses the advertisement entirely. Poisoned reverse (option D) is a stronger variant: instead of suppressing the advertisement, the router advertises the route back with metric = infinity (16 in RIP), explicitly telling the neighbor 'I can't reach D via you'. Poisoned reverse uses more bandwidth but converges faster than simple split horizon.

OSPF router types: A: Internal Router - all interfaces in the same OSPF area; maintains one LSDB. B: ABR (Area Border Router) - has interfaces in multiple areas (including Area 0); maintains separate LSDBs for each area; generates Type 3 summary LSAs to share inter-area routes. C: ASBR (Autonomous System Boundary Router) - connects the OSPF domain to external routing protocols (BGP, RIP, EIGRP, static routes); generates Type 5 external LSAs for redistributed routes. D: DR (Designated Router) - elected on broadcast multi-access networks (Ethernet) to reduce the number of adjacencies. Without DR, n routers would form n(n-1)/2 adjacencies; with DR, each router forms adjacency only with DR and BDR (Backup DR), reducing to n-1 adjacencies. LSAs are flooded through the DR.

Convergence speed comparison: RIP: (1) Waits for the next periodic update (up to 30 seconds) to propagate failure information. (2) Count-to-infinity causes incremental, slow propagation of failure information (bad news travels slowly). (3) Holddown timers add further delay. Total convergence: can take several minutes. OSPF: (1) When a link fails, the router immediately sends a new LSA with the failed link's cost set to infinity (triggered update). (2) LSAs are reliably flooded to ALL routers in the area within seconds. (3) Each router immediately reruns Dijkstra's algorithm upon receiving the LSA update. (4) Convergence time: typically seconds to tens of seconds. A is a metric difference, not convergence. D is reversed (OSPF uses its own reliable flooding mechanism, not TCP or UDP in the traditional sense).

Hot potato routing: a router treats packets like 'hot potatoes' - it wants to get rid of them immediately. The router forwards each packet on the outgoing link with the shortest queue (least congestion), regardless of whether this link is on the optimal path to the destination. The goal is to minimize local queueing delay, not minimize end-to-end path length. This is used in: some ISP intra-AS routing (get traffic off the AS network quickly - 'early exit' or 'closest exit' routing), some optical networks, deflection routing in fault-tolerant systems. Trade-off: reduces local congestion but may increase total path length and end-to-end delay.

ARP request frames are sent using the Ethernet broadcast MAC address in the local subnet.

Autonomous System (AS): a collection of IP networks under a single administrative domain with a consistent routing policy (e.g., an ISP, a university, a company). IGP (Interior Gateway Protocol): routing within a single AS. Examples: OSPF (link-state), RIP (distance vector), EIGRP (hybrid), IS-IS (link-state). IGPs can optimize for performance metrics (hop count, bandwidth, delay). EGP (Exterior Gateway Protocol): routing between different ASes. The only currently used EGP is BGP-4. EGPs must support policy-based routing (business relationships, contracts), AS loop prevention, and scalability to the entire internet (BGP routing tables currently hold ~900,000+ prefixes).

BGP (RFC 4271) is the exterior gateway protocol (EGP) - the only routing protocol used between Autonomous Systems (ASes) on the global internet. BGP is a path vector protocol: each route advertisement includes the full AS path (sequence of AS numbers) to the destination, which: (1) Prevents routing loops - if a router sees its own AS in the path, it rejects the route. (2) Enables policy-based routing - ISPs apply complex policies (prefer certain paths, avoid certain ASes) regardless of shortest path. BGP operates over TCP port 179 (not UDP) for reliable session. BGP metric: LOCAL_PREF, AS_PATH length, MED - not simple hop count.