Routing fundamentals and the routing table

Why this matters

Switching moves frames within a network. Routing moves packets between networks, and it is what makes an internet possible rather than a collection of isolated segments.

Objective 2.1 is "explain characteristics of routing technologies", so the questions are conceptual rather than command-line: what a protocol is for, when you would choose one over another, and why a router picked the path it picked.

The lesson

Static routing

A static route is one you configure by hand: for this destination network, send to this next hop.

Its properties are exactly what you would expect. It is predictable — it does exactly what you wrote and never changes its mind. It consumes no CPU or bandwidth, because nothing is being advertised or recalculated. It reveals nothing about your topology to a neighbour.

And it does not adapt. If the next hop dies, the route stays in the table pointing at a dead path until a human removes it. Every change is manual, which does not scale.

Use static routing for small networks, for stub sites with one way out, and for the default route — 0.0.0.0/0, the "if you match nothing else, send it here" entry that most networks need exactly one of.

Dynamic routing

Dynamic routing protocols let routers advertise what they know and learn from each other. They discover topology automatically, adapt when links fail, and scale to networks nobody could maintain by hand.

The cost is CPU, memory and bandwidth for the protocol itself, plus more moving parts to understand when something goes wrong.

The first division is interior versus exterior. An interior gateway protocol (IGP) routes within one organisation's network — one autonomous system. An exterior gateway protocol (EGP) routes between autonomous systems. That distinction is the one exam questions hang on.

EIGRP

EIGRP is Cisco's interior protocol, an advanced distance-vector design (sometimes called hybrid, because it borrows ideas from link-state).

Its distinguishing features:

• A composite metric based primarily on bandwidth and delay, so it prefers genuinely faster paths rather than merely shorter ones.
• The DUAL algorithm, which precomputes a feasible successor — a guaranteed loop-free backup route. When the primary fails, the backup is already there, so convergence is very fast.
• Sends incremental updates — only what changed — rather than periodic full tables.

It was proprietary for most of its life and was later published as an open standard, but in practice it is still a Cisco-estate protocol. The exam cue is "Cisco environment, fast convergence".

OSPF, in outline

OSPF is the standards-based interior protocol, and a link-state design: every router builds a complete map of the area and runs Dijkstra's algorithm to compute shortest paths. It is covered in full in the next lesson, along with route selection.

The one-line contrast: EIGRP is Cisco and distance-vector; OSPF is open and link-state.

BGP

BGP (Border Gateway Protocol) is the exterior protocol — the one that routes between autonomous systems, and the protocol the public internet runs on.

It is a path-vector protocol: rather than a numeric cost, it carries the full list of autonomous systems a route has traversed. That AS-path is used both for loop detection (a router rejects any route already containing its own AS number) and for policy.

The property that sets BGP apart is that it selects paths by policy, not by speed. An organisation can prefer a route because of cost, contract or politics, and express that preference. On the internet, the shortest path is often not the chosen one, and that is deliberate.

BGP is also slow and deliberate — it does not react instantly to changes, because a global routing table cannot afford to churn.

The exam cue is unambiguous: if a question mentions the internet, multiple ISPs, or autonomous systems, the answer is BGP.

Address translation: NAT

NAT (Network Address Translation) rewrites addresses in the IP header as packets cross a boundary — almost always translating private RFC 1918 addresses to a public one on the way out and reversing it on the way back.

It exists because IPv4 ran out of addresses, and it is why the entire world's home and office networks can use 192.168.x.x simultaneously.

The variants:

• Static NAT. A permanent one-to-one mapping between one private and one public address. Used when a server must be reachable at a fixed public address.
• Dynamic NAT. A pool of public addresses, handed out as needed. Still one-to-one while in use, so it runs out when the pool is exhausted.
• PAT (Port Address Translation), also called NAT overload, maps many private addresses to one public address by tracking source ports. This is what nearly every network actually uses, and it is covered in the next lesson.

Two honest downsides the exam expects you to know. NAT breaks end-to-end addressing, which complicates protocols that embed addresses in their payload — SIP and FTP are the standard examples, which is why firewalls need protocol-aware helpers for them. And NAT is not a security control, though it is often described as one: it hides internal addressing as a side effect, but a stateful firewall is what actually enforces policy.

First hop redundancy

A host has one default gateway. If that router fails, the host is isolated — it has no mechanism to discover an alternative, no matter how many other routers exist on the segment.

FHRP (First Hop Redundancy Protocol) is the family of protocols that solves this. Two or more routers cooperate to present a single virtual gateway:

• A virtual IP (VIP) address, which is what hosts are configured to use as their default gateway.
• A shared virtual MAC address, so ARP entries on the hosts remain valid.
• One router active, forwarding traffic; the others standby, monitoring it with hello messages.

If the active router fails, a standby takes over the virtual IP and virtual MAC within seconds. The hosts never change configuration and never notice — which is the entire point, and the answer the exam wants.

The protocols by name: HSRP (Cisco), VRRP (the open standard), and GLBP (Cisco, which additionally load-balances across the group rather than leaving standby routers idle).

A useful way to hold the distinction: link aggregation gives redundancy for a link, spanning tree gives redundancy for a layer 2 path, and FHRP gives redundancy for a gateway. Questions mix them, and the clue is what failed.