STP varieties

Several varieties of spanning tree exist:

When Cisco documentation says “implementing RSTP,” it means Cisco’s implementation: Rapid PVST+.

Worth memorizing the standards cold: 802.1D versus 802.1w is an easy exam question.

Comparison

Protocol Standard Resources Convergence Trees
STP 802.1D Low Slow One
PVST+ Cisco High Slow One per VLAN
RSTP 802.1w Medium Fast One
Rapid PVST+ Cisco Very high Fast One per VLAN
MSTP 802.1s Medium or high Fast One for multiple VLANs

Slow means 30-50 seconds to converge; the rapid versions bring that down to roughly 4-6 seconds through improved timers and enhanced BPDU exchanges.

The resource column is about instance count. 250 VLANs under PVST+ or Rapid PVST+ means 250 spanning tree instances chewing CPU and memory. MSTP exists for exactly that problem: the administrator maps VLANs onto a small number of instances instead.

Per protocol

STP (802.1D)

One instance means the lowest CPU and memory cost, but also one root bridge and one tree: traffic for all VLANs flows over the same path, which can be suboptimal. Convergence is slow. On a topology change (a link goes down), the switch generates a Topology Change Notification (TCN) BPDU and sends it to the root bridge. The root then notifies all nonroot switches and tells them to drop their MAC address aging timer to 15 seconds.

PVST+

One 802.1D instance per VLAN, which enables per-VLAN root bridges and per-VLAN load balancing: switch A as root for VLAN 10, switch B for VLAN 20. Supports PortFast, UplinkFast, BackboneFast, BPDU guard, BPDU filter, root guard, and loop guard. The topology change mechanism is the same as 802.1D, just per VLAN, so convergence is still slow.

RSTP (802.1w)

Fixes convergence, not traffic flow (still one instance, one tree). On a topology change the switch sends a TCN BPDU to its neighbors and immediately flushes its own MAC address table; each receiving switch repeats this until all have. No waiting on the root bridge before aging out addresses, which is where the speed comes from. CPU and memory sit slightly above original STP but below PVST+.

Blocking is called discarding in rapid spanning tree, and the difference isn’t just the name: a discarding port actually listens and learns from the BPDUs arriving on it, so it can fail over as a hot standby instead of starting cold.

Rapid PVST+

One 802.1w instance per VLAN: fast convergence and per-VLAN load balancing, at the price of the largest CPU and memory requirements of the bunch. Topology change mechanism is RSTP’s, per VLAN.

MSTP (802.1s)

Maps many VLANs with the same physical and logical topology into a common RSTP instance with a common root bridge. Cisco’s implementation provides up to 16 instances of 802.1w. Each instance supports PortFast, BPDU guard, BPDU filter, root guard, and loop guard. Resource cost lands between RSTP and Rapid PVST+. Topology changes work like RSTP but are calculated per MSTI (multiple spanning tree instance).

MSTP is also the interoperability answer: it’s the one open standard spanning tree that Catalyst switches support, so a mixed Cisco and non-Cisco environment runs 802.1s. The catch is that the VLAN-to- instance mapping is manual. MST configuration is CCNP material, not CCNA.

Catalyst defaults