STP (802.1D) vs PVST+ and RSTP (802.1w) vs Rapid PVST

[sergeyugcorp]

In Cisco 3550, the default Spanning Tree Protocol is PVST. The command to configure this protocol is spanning-tree mode pvst. However, when I issue the show spanning-tree command, the output says, Spanning tree enabled protocol ieee. Even though PVST is based on 802.1D, it is clearly not an IEEE standard. Why is it listed as ieee?

Now, another Spanning Tree Protocol option is rapid-pvst. My understanding is that Rapid PVST is a Cisco proprietary protocol, but when I configure spanning-tree mode rapid-pvst and issue the show spanning-tree command, the output says, Spanning tree enabled protocol rstp.

I thought RSTP is an IEEE 802.1w standard (later integrated into 802.1Q-2005), and is not quite the same thing as Cisco's Rapid PVST.

I'm configuring a lab to match our production environment.  I recall reading that Nexus 7K has Rapid PVST running by default, but when I issue the show spanning-tree command In Nexus 7K, the output says, Spanning tree enabled protocol rstp.

Is Cisco intentionally using IEEE names for their proprietary protocols - IEEE for PVST and RSTP for Rapid PVST? Why?

Could someone in the know shed some light on this?

Thanks!

[Peter Paluch]

Hello,

Actually, when talking about PVST+ or Rapid PVST+, it means that each VLAN runs a separate, standalone instance of 802.1D STP (in case of PVST+) or 802.1w RSTP (in case of Rapid PVST+). In other words, each VLAN runs a perfectly IEEE-compliant STP or RSTP inside. If you connect to an access port in a particular VLAN, you will see absolutely normal STP or RSTP BPDUs. By observing BPDUs sent from an access port, you have no way of telling if the switch runs plain STP/RSTP or a Per-VLAN STP/RSTP.

PVST+ or Rapid PVST+ BPDUs are visible only on trunks.

This is how the outputs of show spanning-tree should be interpreted:

VLAN0138

  Spanning tree enabled protocol ieee

  Root ID    Priority    32906

             Address     0009.e8ee.02c0

             Cost        8

             Port        43 (GigabitEthernet0/43)

             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

[ cut ]

Here, the switch claims that the VLAN138 runs "IEEE" STP which is true: within this particular VLAN, STP is run. However, the switch as a whole is configured in PVST mode.

The same switch would say in show spanning-tree summary:

Switch# show span sum

Switch is in pvst mode

Root bridge for: VLAN0026, VLAN0128-VLAN0133, VLAN0135-VLAN0137

[ cut ]

Does this clarify things a little?

Best regards,

Peter

[johnnylingo]

The way I'd put it is the mode and the protocol are two different things.

PVST mode uses 802.1d as the Spanning-Tree protocol.  Cisco switches call this protocol "IEEE"

R-PVST or MST modes use 802.1w as the Spanning-Tree protocol.  Cisco switches call this protocol "rstp"

The 'show spanning-tree summary' command will display the mode.  The 'show spanning-tree' command will show the protocol. 

[Peter Paluch]

Hello Johny,

Thanks for joining. This is a nice view your have presented, and clarifies many things. Thanks!

Two comments, though. PVST or RPVST is not just a mode of running STP or RSTP. PVST and RPVST also includes its own BPDUs carried across trunk ports which use a slightly extended format when compared to IEEE-compliant BPDUs. (R)PVST also includes standalone algorithms to deal with inconsistency issues in case of native VLAN mismatches. Hence, it is a protocol in its own right - it has its own messages and its own algorithms.

Secondly, it is true that RPVST and MST both internally use 802.1w RSTP. However, this statement is indirectly suggesting an overly strong similarity between RPVST and MST while in reality, they're quite different. RPVST simply instantiates RSTP for each VLAN independently, and makes sure that the individual per-VLAN BPDUs are accordingly sent and received via trunk ports. However, there is no further cooperation between individual RSTP instances in VLANs. In MSTP, individual instances do run 802.1w RSTP - or better said, they use RSTP semantics instead of running a standalone RSTP processes - but they are much more interrelated: the timers are inherited from MST instance 0, only a single BPDU is used to carry information about all MST instances, internal non-zero MST instances inherit the role/state of a boundary port as determined by MST instance 0, and some other.

Nevertheless, I find your approach to "pvst/rapid-pvst" as the mode and "ieee/rstp" as the protocol to be very clear and very descriptive, and it removes most of the doubts! Good work!

Best regards,

Peter