50 Expert Q&As + 15-Question Interactive Quiz

variance — configured under the router EIGRP process.

1 — meaning only equal-cost paths are installed by default.

128 — the highest multiplier EIGRP supports.

Successor FD × variance ≥ candidate metric — the candidate path must satisfy this to be installed.

No — variance cannot install a path that fails Feasibility.

maximum-paths — configured under the router EIGRP process.

4 — raised to 16 or 32 in modern IOS versions.

32 — depending on platform and IOS version.

offset-list — increases the metric of specific routes matched by an ACL.

No — offset-lists can only increase metrics.

4 out of 5 — the share count reflects the number of packets per cycle.

256 — outbound offsets are divided by 256 and rounded up on downstream routers.

show ip protocols — displays both values in the EIGRP section.

show ip route [destination] — displays "traffic share count is N" for each path.

Proportional load sharing — the correct technical term (not "load balancing").

Equal-cost load balancing installs paths with identical metrics automatically, while unequal-cost load balancing (variance) allows paths with different metrics to be installed if they satisfy the variance formula and the Feasibility Condition.

Because the Feasibility Condition is the fundamental loop-prevention rule in DUAL, and variance is only a metric-comparison amplifier — it cannot override the loop-free guarantee.

Share count is a ratio that determines how many packets each path receives, calculated as Highest Metric / Path Metric for all installed paths.

An inbound offset-list changes the local metric directly, while an outbound offset-list inflates the metric as advertised to downstream neighbors, with the value divided by 256 and rounded up.

Delay is a logical value that only affects routing calculations, while bandwidth affects QoS policies, queue behavior, and interface operational semantics — making delay the safer tuning knob.

Stateful firewalls, load balancers, and TCP-based applications can fail due to asymmetric routing when forward and return traffic use different-latency paths.

It indicates that out of every sum-of-all-share-counts packets, this path will receive N of them, giving a proportional distribution based on metrics.

Yes — maximum-paths controls how many paths can be installed, and variance controls whether unequal-cost paths can be included in that set.

Too-high variance can install many paths with widely varying metrics, which can lead to packet reordering, asymmetric routing, and unexpected traffic patterns.

The downstream router receives the route with an inflated metric, causing it to prefer a different path — but the offset is scaled by 256, so small values may not have a visible effect.

FD is the metric of the primary successor, and all installed paths' metrics are compared against the FD to compute proportional share counts.

Because maximum-path only installs paths with identical metrics by default; to install paths with different metrics, variance must be configured separately.

Always verify the topology table with show ip eigrp topology after applying an offset-list, and confirm that the backup path's RD remains below the new FD.

Delay is the preferred metric because it is logical and per-interface, allowing precise adjustment without affecting other routers or upper-layer protocols.

Proportional load sharing — the accurate technical term, because traffic is distributed proportionally based on share counts, not evenly.

First Check: Verify the Feasibility Condition on the candidate path.

show ip eigrp topology all-links | include [destination] -A 5

If the candidate's RD is greater than the primary FD, the path fails Feasibility and cannot be installed by variance. You must fix the metric relationship first — usually by reducing delay on the backup path or increasing delay on the primary.

Approach:

1. Enable variance with a value large enough to install both paths
2. Calculate target metrics: for 4:1, the backup path's metric should be roughly 4× the primary's metric
3. Tune delay on the backup path to reach the target metric

interface GigabitEthernet0/2
 delay [value]

router eigrp 1234
 variance 4

Verify with show ip route [destination] and confirm share counts of 4 and 1.

Root Cause: The offset-list inflated the backup path's RD past the FD, breaking Feasibility.

Fix:

! Reduce the offset to a value that keeps RD < FD
router eigrp 1234
 no offset-list MATCH_ROUTE in 50000 Gi0/2
 offset-list MATCH_ROUTE in 1000 Gi0/2

Always calculate the maximum allowable offset as: FD - RD. An offset larger than this will always break Feasibility.

Diagnosis: Variance is causing forward traffic to use Path A and return traffic to use Path B.

Options:

1. Remove variance — falls back to single-path forwarding
2. Make paths equal-cost — tune delay so both paths have identical metrics; then maximum-paths installs them symmetrically
3. Suppress return traffic on one path using an outbound offset-list on the neighbor

The preferred fix in modern networks is option 2 — equal-cost load balancing with symmetric paths.

router eigrp 1234
 maximum-paths 8

Verify with show ip protocols | include Maximum path and show ip route [destination]. Modern IOS allows up to 32 paths, but the exact limit depends on platform and available hardware resources.

Cause: Outbound offsets are divided by 256 and rounded up. An offset of 1000 is interpreted as 4 on the downstream router — too small to affect path selection.

Fix: Use a multiple of 256 as the offset value. For a meaningful +100 downstream effect, use offset 25600 (100 × 256).

router eigrp 1234
 no offset-list MATCH_ROUTE out 1000 Gi0/1
 offset-list MATCH_ROUTE out 25600 Gi0/1

Possible Causes:

• CEF per-packet load balancing is disabled (only per-flow is active), so a single flow always takes one path
• The metrics are actually close enough that both round to share count 1
• You are seeing only one flow's distribution, not aggregate traffic

Verify: Generate multiple flows (different source ports) and re-check share counts. Run show ip route [destination] and look at the actual metrics — share count is calculated from those exact values.

! Step 1: Make sure both paths qualify as successors
! Step 2: Tune delay on primary so it carries more share
interface GigabitEthernet0/1
 delay 10

! Step 3: Set variance to install the backup
router eigrp 1234
 variance 3
 maximum-paths 2

! Expected: primary share count 2, backup share count 1
! (Actually distributes more traffic on the higher-bandwidth link)

Note: For precise bandwidth-matched distribution, delay tuning must be exact. Calculate target metric = primary metric × 2 for a 2:1 split.

Approach: Use an inbound offset-list to inflate the fiber path's metric.

ip access-list standard MATCH_ROUTE
 permit 10.1.10.0 0.0.0.255

router eigrp 1234
 offset-list MATCH_ROUTE in 50000 GigabitEthernet0/1

! Now the fiber path is 50000 more expensive for this route
! WAN path is preferred instead

Warning: Verify Feasibility remains valid. If the inflation breaks Feasibility, the backup becomes unreachable during primary failure.

No. Variance enables more paths to carry traffic, which can increase aggregate throughput, but it does not make EIGRP itself faster. Convergence time and query behavior are unrelated to variance. If anything, too-high variance can slow convergence because more paths need reconverging during failures.

Method:

! Step 1: Baseline interface counters
show interfaces Gi0/1 | include output packets
show interfaces Gi0/2 | include output packets

! Step 2: Generate traffic from Host-A to Host-B
! Step 3: Re-check interface counters after 60 seconds
show interfaces Gi0/1 | include output packets
show interfaces Gi0/2 | include output packets

! Both counters should show traffic — confirming the distribution

Each installed path becomes a CEF adjacency entry with its own load-share factor. With variance, multiple adjacencies exist for the same prefix, and CEF distributes traffic based on per-flow hashing across those adjacencies. Verify with show ip cef [prefix] internal to see the load-share weights.

The maximum ratio depends on the actual metrics. If the primary FD is 100 and the third path is at 400 (which passes the variance 4 check: 100×4=400), the share counts would be:

• Path 1: 400/100 = 4
• Path 2: 400/200 = 2
• Path 3: 400/400 = 1

Total of 7 shares across 3 paths. Ratio is 4:2:1 between primary, secondary, tertiary.

! Before removing, verify the primary path is stable
show ip route [destination]
show ip eigrp topology [destination]

! Then remove variance
router eigrp 1234
 no variance

! Traffic reverts to primary path only
! Backup becomes a passive FS

Note: Removal is non-disruptive for existing flows — EIGRP simply stops installing the secondary path. Existing flows may experience a brief flap as CEF updates.

Cause: Setting variance but forgetting that the path must first qualify as a Feasible Successor. Because variance does not generate syslog messages when it fails to install a path, engineers assume it's working.

Verify: Always check show ip eigrp topology all-links and show ip route [destination] immediately after enabling variance.

No. EIGRP is a control-plane protocol that operates on destination prefixes. It cannot distinguish between traffic types within the same destination. The correct solution is QoS combined with EIGRP — use EIGRP to install multiple paths, and QoS to prioritize traffic on each path. CEF per-flow hashing handles distribution; QoS handles prioritization.

Approach:

1. Apply an inbound offset-list on the unstable path to make its metric just outside the variance range
2. Or, use route filtering (distribute-list with prefix-list) to exclude the unstable path from installing that prefix
3. Or, fix the underlying instability — variance exposes instability that was already present

The preferred fix is to address the root cause, but offset-lists provide an interim control mechanism.

If the secondary path was already installed via variance, it becomes the only remaining path — traffic shifts entirely to it with no convergence delay. This is a key advantage of variance: it provides instant failover because the backup is already in the routing table, not just in the topology table.

Monitoring Strategy:

1. Poll interface counters periodically (SNMP or CLI)
2. Calculate delta per path per interval
3. Compare actual ratio against share-count target
4. Alert if the ratio deviates significantly (indicating path changes)

For automated monitoring, use SNMP OIDs for interface counters and time-series databases (Grafana, PRTG) to visualize the ratio over time.

Approach: Double the metric of the path you want to reduce traffic on.

! Step 1: Baseline current metrics
show ip route [destination]

! Step 2: Increase delay to double the metric
interface GigabitEthernet0/2
 delay [current-delay value x 2]

! Step 3: Verify share count reflects the new metric
show ip route [destination]
! Expected: previous share count on Gi0/2 now halved

Adjust incrementally and verify at each step, because metric changes interact with the FD and Feasibility Condition.