IPv4 Static Routes: Next-Hop and Exit-Interface

A static route is you telling the router, by hand, that to reach this prefix it should go this way. Connected routes cover subnets on the router's own interfaces; everything one or more hops away needs either a static route or a dynamic protocol. Static routes are precise, predictable, and generate no protocol traffic — ideal for stub sites, default routes to an ISP, and small topologies — but every one is hand-maintained, so they do not scale to large or changing networks.

ip route <destination-prefix> <mask> { <next-hop-ip> | <exit-interface> | <exit-interface> <next-hop-ip> }

When you write ip route 10.1.4.0 255.255.255.0 10.1.2.2, the router installs the route — but 10.1.2.2 is not an interface, it is another router's address. To actually forward, the router must answer a second question: how do I reach 10.1.2.2? It searches the table again, finds 10.1.2.2 inside the connected route for 10.1.2.0/30 off Gi0/1, and now knows to send the frame out Gi0/1 and ARP for 10.1.2.2. That two-step match-then-resolve is recursive lookup.

This is where static routing bites people, and it bites twice. Routing decisions are made per-router and per-direction, so a packet only completes a round trip if every router along the way can forward it both out and back.

And even with every router able to forward outbound, the reply has to find its own way home. If the far router has no route back to the sender's subnet, the echo request arrives and the echo reply is dropped. The forward path was flawless; the missing return route killed it. That is why a single ip route command on one router is never enough for a two-way ping.

Topology: R1 — R2 — R3 in a line. R1's LAN is 10.1.1.0/24 off Gi0/0. The R1-to-R2 link is 10.1.2.0/30, with R1 at .1 and R2 at .2. The R2-to-R3 link is 10.1.3.0/30, with R2 at .1 and R3 at .2. R3's LAN is 10.1.4.0/24 off its Gi0/0. The goal is a working ping from a host on 10.1.1.0/24 to 10.1.4.7.

! R1 — reach R3's LAN via R2
R1(config)# ip route 10.1.4.0 255.255.255.0 10.1.2.2

! R2 — the middle router needs BOTH directions
R2(config)# ip route 10.1.4.0 255.255.255.0 10.1.3.2
R2(config)# ip route 10.1.1.0 255.255.255.0 10.1.2.1

! R3 — the return route back to R1's LAN
R3(config)# ip route 10.1.1.0 255.255.255.0 10.1.3.1

R1# show ip route static
      10.0.0.0/8 is variably subnetted, 6 subnets, 3 masks
S        10.1.4.0/24 [1/0] via 10.1.2.2

R1# show ip route 10.1.4.7
Routing entry for 10.1.4.0/24
  Known via "static", distance 1, metric 0
  Routing Descriptor Blocks:
  * 10.1.2.2
      Route metric is 0, traffic share count is 1

Read the installed route: S marks it static, [1/0] is administrative distance 1 and metric 0 — static routes carry AD 1 by default and no meaningful metric — and via 10.1.2.2 names the next hop. show ip route static filters to only the routes you configured, which is the fastest way to audit your own work. show ip route 10.1.4.7 proves the forwarding decision for one specific host.

• ip route <prefix> <mask> then a next hop, an exit interface, or both — three forms of one command.
• Next-hop is the safe default on Ethernet; exit-interface-only suits point-to-point serial; fully specified is most robust when you want to pin the interface.
• Recursive lookup: a next-hop route triggers a second search to resolve the next hop, usually against a connected route.
• Every router on the path needs a route to the destination — including the middle ones, which are not exempt.
• Every forward route needs a matching return route, or the ping fails in one direction only.
• Static routes carry AD 1 and appear as S <prefix> [1/0] via <next-hop>.
• Verify with show ip route static, and show ip route <destination> for the exact winning entry.