Some WAN Carriers make their 'protection' bandwidth available to ISPs and other
takers on a preemtable basis. That is, when a ring failure occurs the protection
bandwidth is reclaimed, or preempted, by the carrier for restoration purposes,
thus "bumping" the original occumpant (you) into the ditch.
Since this type of bandwidth comes at a discount, one has to weigh the
operational risks (SLAs?) against the financial tradeoffs, which is what I
suspect you are doing.
It isn't all that bad if used in a straightforward mesh topology, or one where
one for one link protection is used, because the ISP will normally design to
recover their mesh through regular Layer 3 routing routines, anyway. It can get
out of hand, however, if the particular ring in question is having intermittent
troubles, causing the ISP to encounter an undesirable number of convergences per
unit of time (hours, day, week, etc).