PacketLab
Switching · Spanning Tree

Spanning-tree limitations

Choose Layer2 boundaries with bandwidth, convergence and failure scope in mind.

Independent lab examples; selected output and frame summaries are illustrative. Port naming and command support vary by switch and software release.

Network and objective

Classic PVST+ is used to expose original STP behavior; rapid-PVST+ and MST are separate lessons.

Use three fresh isolated switches with Gigabit trunks: SW1 Gi1/0/23 ↔ SW2 Gi1/0/23; SW1 Gi1/0/24 ↔ SW3 Gi1/0/24; SW2 Gi1/0/24 ↔ SW3 Gi1/0/23. Both VLANs10/20 exist on every switch. Baseline SW1/SW2/SW3 priorities are4096/8192/12288, with short costs and no extra links. Start independently from the included baseline for each lesson; subsequent repair snippets apply to the named device.

Ethernet switchingIllustrative topology
Physical links are solid; device roles and relevant links are labeled.
Physical links are solid; device roles and relevant links are labeled.

How it works

One tree creates a loop-free path but can leave redundant capacity idle for a particular VLAN. Different PVST/MST roots can distribute VLAN groups; they do not deliver arbitrary per-flow ECMP within a single tree. Traffic can follow a root-oriented path that is not the shortest path between a particular pair of hosts.

Large stretched broadcast domains also enlarge flooding and failure scope. Rapid STP improves recovery conditions but does not eliminate control loss, misconfiguration or every outage. Designs can use routed access/fabric ECMP or engineered multichassis aggregation where supported; these introduce their own operational and control-plane requirements.

Configure or inspect

SW1

vlan 10
vlan 20
spanning-tree mode rapid-pvst
spanning-tree pathcost method short
spanning-tree vlan 10 priority 4096
spanning-tree vlan 20 priority 4096
interface range GigabitEthernet1/0/23 - 24
 switchport mode trunk
 switchport trunk allowed vlan 10,20
 switchport nonegotiate
 no shutdown
exit

SW2

vlan 10
vlan 20
spanning-tree mode rapid-pvst
spanning-tree pathcost method short
spanning-tree vlan 10 priority 8192
spanning-tree vlan 20 priority 8192
interface range GigabitEthernet1/0/23 - 24
 switchport mode trunk
 switchport trunk allowed vlan 10,20
 switchport nonegotiate
 no shutdown
exit

SW3

vlan 10
vlan 20
spanning-tree mode rapid-pvst
spanning-tree pathcost method short
spanning-tree vlan 10 priority 12288
spanning-tree vlan 20 priority 12288
interface range GigabitEthernet1/0/23 - 24
 switchport mode trunk
 switchport trunk allowed vlan 10,20
 switchport nonegotiate
 no shutdown
exit

Verify the result

Illustrative output · selected lines
SW3#show spanning-tree vlan 10
Gi1/0/24 Root FWD 4
Gi1/0/23 Altn BLK 4
SW3#show interfaces counters
Port      InOctets OutOctets
Gi1/0/24  1200000  1100000
Gi1/0/23  1200     1200

The blocked port can carry control traffic while ordinary VLAN10 data uses the forwarding path. The numbers are constructed, not a performance benchmark; another VLAN may use a different tree.

Failure and recovery

Do not solve unused bandwidth by disabling STP on a redundant cycle. Decide whether VLAN-group placement, a valid channel, a supported multichassis design or routing is appropriate, and test its failure behavior.

Retain loop control while evaluating an alternative in an isolated lab. Compare permitted forwarding, convergence, operational complexity and failure containment.

Further reading

Key Takeaways

  • A single tree is loop-free rather than arbitrary per-flow ECMP.
  • Root placement can create indirect host-to-host paths.
  • VLAN-group sharing differs from same-flow multipath.
  • Large Layer2 scope increases flooding/failure exposure.
  • Alternatives need their own validated control design.

Spanning-tree limitations Quiz

Six questions · four options each · select one answer per question. Submit all answers to see your score and explanations.

1. Does one tree give arbitrary per-flow ECMP?
2. Can a blocked port still carry control frames?
3. What can PVST root distribution share?
4. What grows with a stretched broadcast domain?
5. Is disabling STP a valid bandwidth fix for a cycle?
6. What must an alternative design include?