The admission control rules for this model are described in RFC 4125. Each CT shares a percentage of the Maximum Reservable Link Bandwidth through the user-configured BC for this CT. The Maximum Reservable Link Bandwidth is the link bandwidth multiplied by the RSVP interface subscription factor.
The sum of all BC values across all CTs does not exceed the Maximum Reservable Link Bandwidth. In other words, the following rule is enforced:
SUM (BCc) =< Max-Reservable-Bandwidth, 0 ≤ c ≤ 7
An LSP of class-type CTc, setup priority p, holding priority h (h=<p), and bandwidth B is admitted into a link if the following condition is satisfied:
B ≤ Unreserved Bandwidth for TE-Class[i]
where TE-Class [i] maps to < CTc, p > in the definition of the TE classes on the node. The bandwidth reservation is effected at the holding priority; that is, in TE-class [j] = <CTc, h>. As such, the reserved bandwidth for CTc and the unreserved bandwidth for the TE classes using CTc are updated as follows:
Reserved(CTc) = Reserved(CTc) + B
Unreserved TE-Class [j] = BCc - SUM (Reserved(CTc,q)) for 0≤ q ≤ h
Unreserved TE-Class [i] = BCc - SUM (Reserved(CTc,q)) for 0≤ q ≤ p
The same is done to update the unreserved bandwidth for any other TE class making use of the same CTc. These new values are advertised to the rest of the network at the next IGP-TE flooding.
When DiffServ is disabled on the node, this model degenerates into a single default CT internally with eight preemption priorities and a non-configurable BC equal to the Maximum Reservable Link Bandwidth. This would behave exactly like CT0 with eight preemption priorities and BC= Maximum Reservable Link Bandwidth if DiffServ was enabled.