I have a PBR service policy on a GRE tunnel interface collecting data off a Ten gigabit interface. When the service policy is applied to the tunnel interface the sh proc cpu history is pegged at 100%. When I remove PBR the process isn't impacted at all. I see this when I push 1Gb/s of data between BERT throughput tester devices. I have three other tunnels and the global ip load share algorithm command configured. Two tunnels go over link 1Gb/s and two tunnels go over link 100Mb/s. Why is the process so problematic and how do I mitigate the flow of traffic? I am running sup72033-adventerprisekb_WAN-mz.122-18.SXF9.