EIGRP & DUAL — The Feasibility Condition

EIGRP uses DUAL (Diffusing Update Algorithm) to select the best path and guarantee fast, loop-free routing per destination inside an autonomous system.

Facts

Advanced Distance Vector (hybrid) → the DUAL algorithm
Protocol 88 in the IP header · Multicast 224.0.0.10 (224.0.0.9 old EIGRP routers)
AD: internal 90 · external 170

Core concepts

Term	Short explanation
Successor	Best route (lowest metric/FD)
Feasible Successor	Backup route, loop-free
Feasible Distance (FD)	Your total metric to the destination
Reported Distance (RD)	The neighbor’s metric to the destination
Feasibility Condition	RD < FD (successor)
Passive state	Stable route, no computation
Active state	DUAL running, queries being sent
Query / Reply	Used in active state
Stuck-in-Active (SIA)	Query-reply timeout
Topology table	All known EIGRP routes
Routing table (RIB)	Only the successor is installed

FD = your total metric to the destination via a neighbor – used to pick the successor and compare against RD
RD (Advertised Distance) = what the neighbor itself says its distance is – always from the neighbor’s perspective

Packet types

Packet	What it does	When used
Hello	Discovers and maintains neighbors	Periodically to keep adjacency alive
Update	Sends routing information (routes, metrics)	At startup, topology changes, partial updates
Query	Asks neighbors for alternative routes	Successor lost and no feasible successor
Reply	Answers a query	When the neighbor has an answer (route or “no route”)
ACK	Acknowledges reliable packets	Confirms Update, Query and Reply

Timers

Timer	Default	Applies to	Function
Hello	5 s (LAN) / 60 s (WAN)	Neighbors	Keeps the adjacency alive
Hold	15 s (LAN) / 180 s (WAN)	Neighbors	How long a neighbor is considered up
Active	3 min (180 s)	Routes	Max time a route may stay active
SIA-Query	90 s	Queries	Time before an “are you alive?” query
SIA-Reply	180 s	Replies	Time before a neighbor is declared SIA
RTO	Dynamic	Reliable packets	Retransmission timeout
RTT	Dynamic	Reliable packets	Measured round-trip time

Roles & the feasibility condition

Successor: best route, lowest FD, installed in the routing table, used for forwarding
Feasible Successor: backup route, already loop-free, usable instantly at failure – lives in the topology table, not the routing table

A neighbor can be an FS when:

RD < FD (successor)

If the condition holds, the neighbor’s path cannot point back at you → loop-free. 👉 EIGRP’s mathematical loop guarantee.

States

Passive: normal, route stable, no queries (“passive = all is well”)
Active: successor gone, no feasible successor → router sends queries and waits for replies → slower convergence
Stuck-in-Active (SIA): replies missing/late – typically bad design, missing stubs, too-large query domains; the neighbor can be dropped

Failure handling – two scenarios

A – FS exists (fastest): successor fails → FS promoted → no queries → fast convergence
B – no FS: successor fails → destination goes active → queries → replies → new successor

Where is what?

Table	Contains
Neighbor table	EIGRP neighbors
Topology table	All routes (successor + FS + candidates)
Routing table	Only the successor

SIA vs dead timer

Point	Hold/Dead timer	SIA
About	Neighbor liveness	Route computation
Trigger	Missing Hello	Missing Reply
Protocol	EIGRP (and OSPF)	EIGRP only
Time frame	15/180 s	~180 s
Fix	Stable link / timers	Stub, smaller query domain

EIGRP Stub

Tells other EIGRP routers: “I am an end station – don’t send me queries for networks I can’t provide alternate paths to.” It limits the query domain, prevents SIA, and improves scalability and stability. Without stubs, queries can spread through the whole network, especially via spoke/access routers.

Advantages & disadvantages (CCNP)

Category	Pros	Cons
Convergence	Very fast via DUAL + FS	Can be slow without FS (queries)
Loop-freedom	Mathematically guaranteed (FC)	Depends on correct design
Updates	Partial and triggered	Queries can stress the network
Load balancing	Unequal-cost via variance	More complex config
Scalability	Good in enterprise	SIA risk in large flat topologies
Design	Simple – no areas	Needs hierarchical design for stability
Vendor support	Strong in Cisco	Limited multi-vendor
Stub	Reduces query domain	Must be configured correctly
Standardization	Partially open	Not a full industry standard like OSPF
Failure handling	Fast local failover with FS	SIA on missing replies
Implementation	Easy to deploy and change	Harder to troubleshoot (DUAL)

Configuration

router eigrp 100
 network 192.168.1.0 0.0.0.255
 network 192.168.2.0 0.0.0.255
 no auto-summary
 eigrp router-id 1.1.1.1
 passive-interface default
 no passive-interface GigabitEthernet0/0
!
eigrp stub                     ! variants: connected / connected summary / connected static
!
interface GigabitEthernet0/0
 ip hello-interval eigrp 100 5
 ip hold-time eigrp 100 15

Maintenance & troubleshooting

show ip eigrp neighbors
show ip route eigrp
show ip eigrp topology
show ip eigrp topology all-links
show ip protocols
show ip eigrp topology active
show ip eigrp neighbors detail
show ip eigrp traffic
debug eigrp packets          ! use with care
debug eigrp fsm
undebug all

Filtering (CCNP important)

A) distribute-list + ACL:

access-list 10 permit 10.10.0.0 0.0.255.255
router eigrp 100
 distribute-list 10 in GigabitEthernet0/0
! in = filter what you receive, out = filter what you advertise

B) prefix-list (best practice):

ip prefix-list EIGRP-FILTER permit 10.10.0.0/16 le 24
router eigrp 100
 distribute-list prefix EIGRP-FILTER out GigabitEthernet0/0

C) route-map (most flexible):

ip prefix-list ALLOW permit 10.10.0.0/16
route-map EIGRP-FILTER permit 10
 match ip address prefix-list ALLOW
router eigrp 100
 distribute-list route-map EIGRP-FILTER out GigabitEthernet0/0

D) unequal-cost load balancing:

router eigrp 100
 variance 2
! only feasible successors are used, and they must still satisfy the feasibility condition

Tagged ccna, ccnp, eigrp