Routing Metrics – Concepts, Calculations, and Configuration

1. What Is a Routing Metric?

A routing metric is a numerical value that a routing protocol assigns to each route it knows about. It represents the “cost” of using that path to reach a destination network. When a router learns multiple paths to the same destination from the same routing protocol, it compares their metrics and installs the route with the lowest metric in the routing table.

Each protocol calculates its metric differently — RIP counts routers, OSPF measures bandwidth, EIGRP weighs bandwidth and delay together, and BGP uses a policy-driven set of path attributes. Because the scales are incompatible, metrics are only meaningful within the same protocol. When two different protocols both know a route, the router uses Administrative Distance (AD) to decide which protocol to trust first, and only then does metric come into play.

  Two paths to 10.1.0.0/24 via OSPF:

  Path A: R1 → R2 → R4   OSPF cost = 20  ← INSTALLED (lower)
  Path B: R1 → R3 → R4   OSPF cost = 30  ← kept as backup

  Same destination, OSPF vs RIP:
  OSPF AD = 110  ← INSTALLED (lower AD wins; metric not compared)
  RIP  AD = 120
            

Related pages: Administrative Distance | Route Sources | OSPF Overview | EIGRP Overview | BGP Overview | RIP Concepts | Default Routes | show ip route | show ip protocols | OSPF Configuration | EIGRP Configuration

2. Metric vs. Administrative Distance

These two values are often confused. The key distinction is the scope at which each one operates:

See also: Administrative Distance – Full Guide

3. How Routers Use Metrics to Select the Best Path

The full route selection decision sequence a Cisco router follows is:

1. Longest prefix match — the most specific route always wins first, regardless of metric or AD (e.g., /30 beats /24 for a matching address).
2. Administrative Distance — if the same prefix is known by multiple protocols or sources, the one with the lowest AD is preferred. All others are discarded from the routing table.
3. Metric — once the winning protocol is chosen, the route(s) with the lowest metric from that protocol are installed.
4. ECMP — if two or more routes share the exact same metric (and protocol), all are installed and traffic is load-balanced across them.

  Routing table decision flow for prefix 10.0.0.0/24:

  Source         AD    Metric   Action
  ──────────     ──    ──────   ──────────────────────────────
  Connected       0      0      Would win (if directly connected)
  OSPF           110     20     ← WINS (lowest AD among known protocols)
  RIP            120      4     Discarded (higher AD)

  If two OSPF paths exist:
  OSPF via R2    110     20     ← INSTALLED
  OSPF via R3    110     30     Kept in OSPF topology, not in routing table
            

4. RIP – Hop Count

RIP (Routing Information Protocol) uses the simplest possible metric: hop count, the number of routers that a packet must traverse to reach the destination.

  Destination: 10.2.0.0/24

  R1 → R2 → R3 → R4 → R5   RIP metric = 4 hops  ← preferred
  R1 → R6 → R5              RIP metric = 2 hops  ← actually preferred!

  (Even if R6 is a slow 56k link — RIP does not know or care)
            

See also: RIP Concepts | RIP Configuration

5. OSPF – Cost

OSPF uses cost as its metric, calculated from interface bandwidth. Unlike RIP, OSPF favours higher-bandwidth paths because higher bandwidth produces a lower cost. See: OSPF Overview

! Common OSPF cost values at default reference (100 Mbps)
! Serial (1.544 Mbps) → cost = 100 / 1.544 = 64
! FastEthernet (100 Mbps) → cost = 100 / 100 = 1
! GigabitEthernet (1000 Mbps) → cost = 100 / 1000 = 0.1 → rounded to 1
! 10 GigabitEthernet (10000 Mbps) → cost = 100 / 10000 = 0.01 → rounded to 1

! Fix: raise reference bandwidth for modern high-speed networks
router ospf 1
 auto-cost reference-bandwidth 10000   ! 10 Gbps reference

! Override cost on a specific interface
interface GigabitEthernet0/1
 ip ospf cost 5

! Verify per-interface OSPF cost
Router# show ip ospf interface GigabitEthernet0/1
            

See also: OSPF Single-Area Configuration | OSPF Areas & LSAs | Step-by-Step: OSPF

6. EIGRP – Composite Metric

EIGRP uses a composite metric that can incorporate up to five components, though in practice only bandwidth and delay are used by default (K1 = 1, K3 = 1; K2, K4, K5 = 0).

Classic EIGRP metric formula (default K-values):

Metric = [ (10^7 / min-bandwidth-kbps) + (sum-of-delays / 10) ] × 256

Example:
  Path has minimum bandwidth = 100,000 kbps (100 Mbps FastEthernet)
  Cumulative delay = 200 (in units of tens of microseconds = 2000 µs total)

  Metric = [ (10,000,000 / 100,000) + (200 / 10) ] × 256
         = [ 100 + 20 ] × 256
         = 120 × 256
         = 30,720
            

! View EIGRP topology table (shows composite metric and component values)
Router# show ip eigrp topology
Router# show ip eigrp topology 10.1.0.0/24

! Modify delay on an interface (in tens of microseconds)
interface GigabitEthernet0/1
 delay 100

! Modify bandwidth (affects metric calculation; does not change actual speed)
interface GigabitEthernet0/1
 bandwidth 500000
            

See also: EIGRP Overview | EIGRP Configuration | Step-by-Step: EIGRP

7. BGP – Path Attributes

BGP does not use a single numeric metric. Instead, it uses a set of path attributes that are evaluated in a specific order (the BGP best-path selection algorithm). BGP is primarily policy-driven rather than topology-driven. See: BGP Overview

Exam tip: The BGP best-path selection order is evaluated top-to-bottom; the first attribute that differs between two paths determines the winner. The common mnemonic is N WLLA OMNI (Neighbour Weight, LOCAL_PREF, Locally originated, AS_PATH, Origin, MED, eBGP vs iBGP, IGP metric, Neighbour ID).

8. Protocol Metric Comparison

9. Equal-Cost Multi-Path (ECMP) Routing

When a routing protocol finds two or more paths to the same destination with identical metrics, it installs all of them in the routing table simultaneously. The router then distributes traffic across these paths — a feature called ECMP (Equal-Cost Multi-Path).

! OSPF — increase ECMP paths
router ospf 1
 maximum-paths 8

! EIGRP — equal-cost load balancing (4 paths)
router eigrp 1
 maximum-paths 4

! EIGRP — unequal-cost load balancing with variance
! Routes with metric ≤ (best-metric × variance) are included
router eigrp 1
 variance 2    ! Allow paths up to 2x the best metric

! Verify ECMP in routing table
Router# show ip route 10.1.0.0
            

10. Metric Manipulation and Tuning

Administrators manipulate metrics to force traffic down preferred paths, implement primary/backup topologies, or satisfy business policies that the automatic metric calculation would not produce.

10a. OSPF Cost Tuning

! Set cost explicitly on an interface (overrides calculated value)
interface GigabitEthernet0/1
 ip ospf cost 5

! Raise reference bandwidth so GigE ≠ 10GigE in cost
router ospf 1
 auto-cost reference-bandwidth 10000   ! 10 Gbps = cost 1; 1 Gbps = cost 10
            

10b. EIGRP Bandwidth and Delay Tuning

! Lower bandwidth statement raises EIGRP metric → makes path less preferred
interface Serial0/0
 bandwidth 512          ! in kbps; does NOT change actual clock rate

! Raise delay → makes path less preferred
interface GigabitEthernet0/1
 delay 200              ! in tens of microseconds
            

10c. RIP Offset-List

! Add 3 hops to all RIP routes learned on Gi0/1 → makes path less preferred
router rip
 offset-list 0 in 3 GigabitEthernet0/1
            

10d. Using Route Maps to Set Metrics

! Set metric to 20 for specific prefixes during redistribution
route-map SET_METRIC permit 10
 match ip address ACL_PREFIXES
 set metric 20

router rip
 redistribute ospf 1 metric 5 route-map SET_METRIC
            

11. Metric During Route Redistribution

When routes are redistributed between different protocols, the source protocol’s metric has no meaning in the destination protocol. A seed metric must be defined or IOS will apply a default that may cause the redistributed routes to be unreachable.

! Redistribute OSPF into EIGRP with explicit seed metric
router eigrp 1
 redistribute ospf 1 metric 10000 100 255 1 1500
!                             BW    DLY REL LD MTU

! Redistribute EIGRP into OSPF with seed cost
router ospf 1
 redistribute eigrp 1 subnets metric 30 metric-type 1

! Redistribute OSPF into RIP with seed hop count
router rip
 redistribute ospf 1 metric 3
            

See also: Step-by-Step: Route Redistribution | Default Routes | Floating Static Routes

12. Metrics and Routing Convergence

Metric stability directly affects how quickly and smoothly a network converges after a topology change:

• Stable metrics (e.g., OSPF cost based on interface bandwidth) produce consistent routing tables and fast convergence because the best path does not change unless the topology actually changes.
• Fluctuating metrics (e.g., EIGRP with K2/K4 load or reliability enabled) cause continuous route changes even under steady traffic loads, leading to route flapping and instability. This is why K2 and K4 are disabled by default.
• RIP’s slow convergence is partly a metric problem: because bad news propagates one hop at a time (counting to infinity), a 15-hop route can take minutes to be declared unreachable after a failure. RIP uses split horizon, route poisoning, and holddown timers to mitigate this.
• OSPF and EIGRP use topology databases (LSDB and topology table) to recalculate metrics after failures, converging in seconds rather than minutes.

13. Verification Commands

! Reading the routing table metric field
! Format: [AD/metric]
Router# show ip route
O    10.1.0.0/24 [110/20] via 192.168.1.2, 00:05:12, GigabitEthernet0/1
!                 ^^^ ^^
!                 AD  metric (OSPF cost)

D    10.2.0.0/24 [90/30720] via 192.168.2.2, 00:04:55, GigabitEthernet0/2
!                ^^  ^^^^^
!                AD  metric (EIGRP composite)

R    10.3.0.0/24 [120/3] via 192.168.3.2, 00:01:02, Serial0/0
!                ^^^ ^
!                AD  metric (RIP hop count)
            

14. Best Practices for Metric Design

• OSPF: Always set auto-cost reference-bandwidth to match the fastest links in your network. The default 100 Mbps reference makes GigE and 10 GigE indistinguishable (both cost 1).
• EIGRP: Never enable K2 (load) or K4 (reliability) in production — they fluctuate dynamically and cause route instability. Stick to K1 (bandwidth) and K3 (delay).
• Redistribution: Always define an explicit seed metric when redistributing into RIP or EIGRP — both default to infinity (unreachable) if no metric is provided.
• ECMP: Verify that ECMP paths are truly equal in capacity before enabling load balancing; unequal physical links with equal metrics overload the lower-capacity path.
• Documentation: Record any manual metric overrides (interface cost, bandwidth, delay statements) in network diagrams and change logs — they are not obvious from a plain show ip route.
• Avoid metric-based tricks for failover in OSPF when a simple ip ospf cost on the standby link suffices; keep configurations readable and auditable.
• BGP: Prefer LOCAL_PREF for outbound traffic control within an AS and MED for signalling preferred ingress points to external peers; avoid relying on AS_PATH prepending as a primary design tool.

15. Key Points & CCNA Exam Tips

• Metric = cost of a route within one protocol; AD = trustworthiness of a protocol between protocols. AD is evaluated first.
• Lower metric = more preferred route (same rule as AD)
• RIP metric = hop count; max 15; 16 = unreachable; no bandwidth awareness
• OSPF metric = cost = Reference BW ÷ Interface BW; lower bandwidth → higher cost
• Default OSPF reference bandwidth = 100 Mbps; raise with auto-cost reference-bandwidth on all routers in the domain
• EIGRP composite metric uses bandwidth and delay by default (K1=1, K3=1); formula: [(10⁷/BW) + (delay/10)] × 256
• BGP uses path attributes (Weight → LOCAL_PREF → AS_PATH → Origin → MED …); higher Weight/LOCAL_PREF preferred; shorter AS_PATH preferred
• Metrics are not comparable across protocols — never compare an OSPF cost to a RIP hop count
• ECMP: when multiple paths share the same metric, all are installed and traffic is load-balanced
• EIGRP supports unequal-cost load balancing via the variance command — unique among IGPs
• Redistribution requires a seed metric; RIP and EIGRP default to unreachable if none is specified
• Read routing table metrics as [AD/metric]: [110/20] = OSPF (AD 110), cost 20

Routing Metric Quiz