RIB, FIB, and Forwarding Decisions

Why routing tables matter on H12-811

HCIA-Datacom moves from Layer 2 switching into IP routing: how a router or Layer 3 switch decides the next hop and outbound interface for a packet whose destination is not a connected host MAC on a local segment. Exam items rarely ask philosophy; they ask you to interpret a routing table, explain why one route beat another, and predict what happens when a next hop disappears.

This section builds the mental model used for the rest of the IP routing and OSPF chapters: control plane learns and selects routes; data plane forwards packets. On Huawei literature and HCIA materials you will see RIB and FIB language even when day-to-day CLI shows a single display ip routing-table view. Treat RIB as “what the router knows and prefers” and FIB as “what the hardware or forwarding engine actually uses after installation.”

Control plane vs data plane (exam-friendly view)

When you type a static route or when OSPF finishes SPF, you are changing control-plane state. Successful selection then installs a forwarding entry so the next customer packet does not require a full software route calculation per packet on modern platforms.

You do not need vendor micro-architecture depth on H12-811. You do need this chain:

1. Sources inject candidate routes.
2. The router selects among candidates using match rules (longest prefix, then preference/cost as applicable).
3. Selected routes appear in display ip routing-table (and related views).
4. Forwarding uses those selected paths; verification with ping/tracert confirms end-to-end behavior.

RIB: the IP routing table you study for HCIA

The Routing Information Base (RIB) is the collection of routing information maintained by the IP routing process. In practical HCIA language, the IP routing table is the RIB view candidates stare at for hours:

<Huawei> display ip routing-table

A healthy table always contains Direct (connected) routes for networks configured on up interfaces, often a Local host route for the interface IP itself, plus any static or dynamic routes that won selection. Without connected routes, static next hops that rely on those interfaces cannot resolve, and local services on that subnet break.

Reading the columns that decide exam answers

Exact column labels vary slightly by VRP version, but HCIA expects you to interpret this family of fields:

Proto tells you who offered the route. Pre tells you how trusted that source is by default. Cost is not the same as preference: cost compares routes from comparable selection contexts (for example two OSPF paths), while preference compares different sources to the same prefix.

Filter views you will use constantly:

<Huawei> display ip routing-table
<Huawei> display ip routing-table protocol static
<Huawei> display ip routing-table protocol direct
<Huawei> display ip routing-table 192.168.10.0

protocol static isolates operator-configured static routes—perfect when troubleshooting “I typed the static but traffic still goes via OSPF.” Looking up a specific destination shows which prefix actually matches that IP under longest-match rules.

FIB: forwarding after selection

The Forwarding Information Base (FIB) is the optimized structure the data plane consults to forward packets. After the RIB selects a best path for a prefix, the device installs a corresponding forwarding entry (conceptually into the FIB). On many platforms engineers also say “FIB” when discussing hardware forwarding tables; for HCIA, remember:

• RIB selection problems → wrong or missing routes in display ip routing-table.
• Forwarding/adjacency problems → route exists, but ARP/ND for the next hop fails, interface is down, or policy drops traffic.

A route can exist in the RIB discussion yet still fail to forward if the outbound interface is down, the next hop is unreachable, or a security feature (ACL, blackhole, URPF-style behaviors on some platforms) discards the packet. Always separate “Is the route selected?” from “Can Layer 2 reach the next hop?”

Local delivery vs transit routing

Campus designs often combine VLANIF gateways (connected routes for user subnets) with static or OSPF routes toward the core. The RIB glues those pieces together.

Longest-prefix match: the first rule of destination selection

When a packet arrives with destination D, the router does not pick “any route that includes D.” It picks the route whose prefix covers D and has the longest mask (most specific network). Only after that prefix is chosen do preference and cost matter among competing routes to that same prefix.

Example prefixes in the table:

Packet to 10.1.1.50 uses 10.1.1.0/24 even if a default route or a summary has a better (lower) preference. Specificity beats preference. That single sentence eliminates many wrong exam answers.

Preference vs longest match (do not reverse them)

If Static 10.1.1.0/24 Pre 60 and OSPF 10.1.0.0/16 Pre 10 both exist, traffic to 10.1.1.50 still follows the /24 because the match is longer—even though OSPF’s preference number is “better.” Preference never overrides a more specific prefix.

Recursive next-hop resolution

A recursive route is one whose next hop is not itself a connected address requiring only ARP on a known interface without further IP lookup—commonly a static route that points to a next-hop IP that must be resolved through another route.

Example:

ip route-static 172.16.0.0 255.255.0.0 10.0.0.2

For this to work, the router must know how to reach 10.0.0.2—typically via a Direct route on a connected interface. If 10.0.0.2 is only reachable through yet another static or dynamic route, resolution continues until a connected exit is found or resolution fails.

Exam traps:

• Configuring a static to a next hop on a subnet that is not in the table.
• Pointing a static to a next hop learned only via a route that depends on the static (recursion loop / unusable route).
• Using an interface-only static on Ethernet without understanding ARP behavior (next section covers static forms).

If display ip routing-table shows the static missing or inactive, ask: Is the next hop resolved? Is the outbound interface up?

How routes enter the RIB (sources you must recognize)

Connected routes are not “optional decoration.” They are how the router knows which prefixes are on-link. Static and OSPF build remote reachability on top of that foundation.

End-to-end packet walk (RIB in action)

1. Packet arrives on GE0/0/1; destination IP is 192.168.50.10.
2. Device checks whether destination is local/connected; if not, performs longest-prefix lookup in the routing table.
3. Match hits 192.168.50.0/24 via next hop 10.10.10.1 out GE0/0/2 (selected static or OSPF path).
4. Data plane resolves Layer 2 adjacency for 10.10.10.1 on GE0/0/2 (ARP on IPv4).
5. Frame is rewritten toward the next hop and sent; TTL decrements on the routed hop.

If step 3 fails → no route / default missing. If step 4 fails → route looks fine in RIB but pings fail with encapsulation/ARP symptoms.

Verification habits that score points

Exam traps for RIB/FIB thinking

• Treating preference as more important than longest match.
• Confusing cost/metric with preference.
• Believing the RIB lists every protocol’s candidates equally as active paths—selection leaves one (or ECMP set) installed for forwarding.
• Assuming a displayed route guarantees ARP success to the next hop.
• Mixing mask (routing table, static config) with wildcard (some ACL/OSPF network statements)—wrong mask length changes the prefix entirely.

What “done” looks like for this section

You can open display ip routing-table, identify connected versus static versus OSPF entries, explain why a packet matches a /24 instead of a default, and describe preference as a same-prefix tie-breaker—not a global “always use OSPF” switch. That model is the foundation for static routes, preference design, floating backups, and OSPF in the following sections and chapter.