Thanks Giuseppe.

Ok, but what happens on a ring with 20 switches or more when the root fails ? I saw an RSTP ring built with more than 20 non-Cisco industrial switches that has some troubles converging in case of failure of the root bridge.

There must be a "theoretical" and a "best of practice" limit in a RSTP ring topology....

Hello Gilberto,

if you have a ring topology and the root bridge fails : a new root bridge has to be elected.

With traditional 802.1D without max-age correct settings problem can arise.

With RSTP 802.1W this timer shouldn't play a role.

see

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_white_paper09186a0080094cfa.shtml

the document says that Cisco has implemented a proprietary version of the synchronization process: a simplified version where only the root port is object to handshake.

Also a switch is authorized to accept inferior BPDUs when it loses its connection to the Root Bridge.

As I wrote in my first post you can choice an appropriate value for max-age and other timers that reflects the topology.

if the root bridge fails the ring is broken and all devices have to elect a new root bridge.

Hope to help

Giuseppe

Giuseppe,

thanks a lot for your answer,

I found the attache paper about the rstp convergence problems, which is also referenced in some other writings.

Hello Gilberto,

thanks for the link you have provided in your post.

Even if these are only simulations with a known network modeling tool they show the need for changing the max-age value.

Let us be practical increasing the max-age has no impact during normal and small failures.

Best Regards

Giuseppe

Have a look at the "Characterization of the EttF Cell/Area Zone Design" section of the Ethernet-to-the-Factory design guide. There are also some results of STP loop and convergence tests (up to a 16-switch ring)

http://www.cisco.com/en/US/docs/solutions/Verticals/EttF/ApA_EttF.html