STP vs RSTP: Convergence Comparison

Spanning Tree Protocol (STP, IEEE 802.1D) and Rapid Spanning Tree Protocol (RSTP, IEEE 802.1w) are Layer 2 protocols that prevent loops in Ethernet networks with redundant links. Without them, broadcast storms, multiple frame copies, and MAC table instability would cripple a switched network. Both protocols create a loop-free logical topology by blocking certain ports while keeping all physical links available for redundancy. The fundamental difference is convergence speed: STP relies on timers (up to 50 seconds), while RSTP uses active handshakes and port role assignments to converge in a few seconds (often under 2 seconds).

STP elects a single root bridge based on the lowest bridge ID (priority + MAC address). Each non-root bridge selects one root port (the port with the lowest path cost to the root). On each segment (collision domain), the bridge with the lowest path cost to the root becomes the designated port; all other ports on that segment become non-designated (blocked). The process uses Bridge Protocol Data Units (BPDUs) sent every 2 seconds (Hello timer) by the root bridge, relayed by other bridges.

STP port states: - Blocking: Port is administratively down or loop prevention; no data forwarding, no MAC learning; receives BPDUs but does not send them (except to indicate presence). - Listening: Port transitions from blocking; listens for BPDUs to determine its role; no data forwarding, no MAC learning; duration = Forward Delay (15 seconds by default). - Learning: Port learns MAC addresses from incoming frames but still does not forward data; duration = Forward Delay (15 seconds). - Forwarding: Port forwards data and learns MAC addresses normally. - Disabled: Port is administratively shutdown.

Timers: - Hello: 2 seconds (default) – interval between BPDUs sent by root. - Forward Delay: 15 seconds (default) – time spent in listening and learning states. - Max Age: 20 seconds (default) – time a bridge waits without hearing a BPDU before re-evaluating root.

Convergence: When a topology change occurs (e.g., link failure), the bridge detecting the change sends a Topology Change Notification (TCN) BPDU toward the root. The root acknowledges and sets the Topology Change flag in its BPDUs for a period (Forward Delay + Max Age). All bridges then shorten their MAC address aging time from 300 seconds to Forward Delay seconds (15). Meanwhile, the affected ports go through blocking -> listening (15s) -> learning (15s) -> forwarding. Total convergence can be up to 50 seconds (20s Max Age + 15s listening + 15s learning).

RSTP (802.1w) maintains the same root election and path cost logic but changes port roles and states. RSTP port roles: - Root Port: Best path to root bridge. - Designated Port: Best path from a segment to the root; forwards traffic. - Alternate Port: Backup to the root port; provides an alternate path to the root; immediately transitions to forwarding if the root port fails (without timers). - Backup Port: Backup to a designated port on the same segment (rare in switched networks); provides redundancy for the designated port.

RSTP port states are reduced to three operational states: Discarding, Learning, and Forwarding. The old blocking, listening, and disabled states are all condensed into Discarding.

RSTP convergence uses a proposal-agreement handshake between bridges. When a port is elected as a designated port, it sends a proposal BPDU (with the proposal flag set) to the neighbor. The neighbor, if it agrees, sets its port to the root port and sends an agreement BPDU back. This handshake propagates rapidly from the root outward, typically converging in a few seconds. Edge ports (ports connected to end hosts) can be configured to transition directly to forwarding without handshake, using the spanning-tree portfast command.

RSTP also handles topology changes differently: a bridge detecting a link failure immediately sends a BPDU with the topology change flag set, and all bridges age out MAC entries quickly (topology change timer = 2 x Hello time, typically 4 seconds).

STP defaults:

RSTP defaults:

Switch# show spanning-tree
VLAN0001
  Spanning tree enabled protocol ieee
  Root ID    Priority    32769
             Address     0011.2233.4455
             Cost        19
             Port        1 (GigabitEthernet0/1)
             Hello Time  2 sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    32769  (priority 32768 sys-id-ext 1)
             Address     0011.2233.4466
             Hello Time  2 sec  Max Age 20 sec  Forward Delay 15 sec
             Aging Time  300 sec

Interface        Role Sts Cost      Prio.Nbr Type
---------------- ---- --- --------- -------- --------------------------------
Gi0/1            Root FWD 19        128.1    P2p
Gi0/2            Desg FWD 19        128.2    P2p
Gi0/3            Altn BLK 19        128.3    P2p

For RSTP, the output shows "protocol ieee" but the port states reflect RSTP roles (Altn, Back, etc.). To verify RSTP is active, use:

Switch# show spanning-tree detail
 VLAN0001
   Spanning tree enabled protocol rstp

RSTP is backward-compatible with STP: if a port receives an STP BPDU, it falls back to STP mode on that port (using STP timers). This can cause suboptimal convergence if a legacy switch is in the network. Also, RSTP interacts with VLANs via PVST+ (Per-VLAN Spanning Tree) or Rapid PVST+. In Cisco switches, Rapid PVST+ runs an instance of RSTP per VLAN, which is the default on newer switches. MST (Multiple Spanning Tree) uses RSTP for rapid convergence within instances.

In enterprise networks, the choice between STP and RSTP directly impacts application availability. Consider a campus network with redundant distribution switches. With classic STP, a link failure between an access switch and a distribution switch causes 30-50 seconds of convergence – enough to drop VoIP calls, disrupt database transactions, and cause TCP timeouts. In one real scenario, a hospital network using STP experienced a 45-second outage during a link failure, affecting patient monitoring systems. After migrating to RSTP, convergence dropped to under 2 seconds.

Another common deployment is in data center topologies. Many data centers use RSTP (or MST with RSTP) for server access layer redundancy. For example, a top-of-rack (ToR) switch connecting to two leaf switches. With RSTP, the alternate port immediately takes over if the root port fails, providing sub-second failover. Network engineers often combine RSTP with features like UplinkFast (Cisco proprietary, now obsolete in RSTP) and BackboneFast (also obsolete in RSTP) – but these are unnecessary with proper RSTP.

A frequent misconfiguration is forgetting to set PortFast on edge ports. An engineer might connect a server to a switch port and wonder why it takes 30 seconds to get an IP address via DHCP. The fix is spanning-tree portfast on that port. Also, mixing STP and RSTP in the same network causes all switches to run STP on the mixed segment, negating RSTP's speed. In large networks, engineers must ensure all switches run Rapid PVST+ or MST.

Performance considerations: RSTP scales well up to hundreds of switches per VLAN, but the proposal-agreement handshake can cause CPU spikes during reconvergence in very large topologies. MST (Multiple Spanning Tree) reduces the number of instances, improving scalability. When misconfigured (e.g., inconsistent port roles due to incorrect path costs), RSTP can still converge quickly but may create temporary loops if BPDUs are lost – though the handshake minimizes this risk.