If a router receives a packet and finds no matching route in its routing table, it checks for a default route, also known as the gateway of last resort. If a default route exists, the packet is forwarded to the specified next-hop router associated with it. This allows packets to continue toward their destination, even if the current router does not have a specific route entry for that destination IP address. However, if no matching route or default route exists, the packet is dropped. When this happens, the router may send back an ICMP (Internet Control Message Protocol) message, typically a "Destination Unreachable" message, to the sender, informing them that the delivery failed. This mechanism ensures that undeliverable packets do not endlessly circulate the network. For networks with sensitive information or security restrictions, routers may be configured not to send such ICMP responses to avoid revealing internal routing structure to potential attackers.

Routing loops occur when packets are forwarded in a circular path between routers, never reaching their destination. They usually result from incorrect or outdated routing table information, especially during the convergence period after a topology change. In distance-vector protocols like RIP, loops can easily form due to slow propagation of updates and limited loop prevention mechanisms. To prevent loops, several strategies are employed. Time To Live (TTL) fields in IP packets are decremented at each hop; when TTL reaches zero, the packet is discarded, preventing indefinite looping. In protocols like BGP, AS path checking prevents a router from accepting a route that includes its own AS number, breaking potential loops between autonomous systems. Link-state protocols like OSPF use a global view of the network topology and maintain synchronised databases, reducing loop formation. Route invalidation timers, hold-down timers, and split horizon rules further help to detect and eliminate routing loops efficiently.

When a router identifies multiple equal-cost paths (paths with the same metric value) to the same destination, it can use Equal-Cost Multi-Path (ECMP) routing. ECMP is supported by many modern dynamic routing protocols such as OSPF and EIGRP. It allows a router to install all equal-cost routes in its routing table and distribute traffic across them. The router may use round-robin, per-packet, or per-flow load balancing techniques. Per-packet load balancing sends each packet down a different path in rotation, which may result in packet reordering and affect time-sensitive applications. Per-flow load balancing uses fields in the packet (like source/destination IP or port) to ensure that all packets of a flow use the same path, preserving packet order. ECMP improves bandwidth utilisation, enhances fault tolerance, and provides redundancy. If one path fails, the router can continue to forward packets using the remaining paths without waiting for full convergence.

Convergence time is the period between a change in the network (such as a link going down) and the moment when all routers update their routing tables to reflect that change. Short convergence times are critical for ensuring high availability and minimal packet loss during network disruptions. If convergence is slow, packets may be dropped or misrouted, causing delays or network outages. Different routing protocols have varying convergence characteristics. RIP, for example, has a slow convergence due to its periodic update intervals and basic algorithm. In contrast, OSPF and EIGRP converge more rapidly due to their use of event-driven updates and sophisticated algorithms that calculate shortest paths immediately. BGP has slower convergence by design, especially between autonomous systems, to avoid frequent route flapping and ensure global stability. Network administrators must balance convergence speed with stability; too-rapid convergence can cause oscillations, while slow convergence can degrade user experience and application performance.

Hop-by-hop forwarding involves routers making independent decisions to forward packets one step at a time towards their destination. Each router examines the packet’s destination IP address, consults its routing table, and determines the next best router (next hop). The packet continues this process through several routers until it reaches the destination. This method does not require the source to know the full route, allowing the network to adapt if a router or link fails. Dynamic routing ensures alternate paths are used, maintaining reliability. Hop-by-hop forwarding supports scalability and decentralised management of global Internet routing.