To tear down a connection 
, a teardown request message is sent
along 's path which instructs all SECs and HECs to update the list of
established connections and their blocking guard values.
Unfortunately, a
connection 
admitted at switch 
after , may depend on
the blocking guard value reduced due to the active connection
. We then say that is critical to .
Though ideally we should be able to increment the blocking guard to reflect
the teardown of , we cannot do so until all connections for which
was critical are torn down.
On receiving the teardown request message, the SEC at switch
marks as being torn down and checks to see if there are active
connections for which was critical. If not, it increments the
blocking guard value 
to reflect the teardown of .
The SEC also recursively completes updating other blocking guards at switch
, to reflect previously torndown connections which were critical
to . This was delayed until now because of being active.
The teardown procedure is complete after the teardown request message reaches the destination host.