It seems that when abrt detects crashes, it generates traffic to Satellite 6 candlepin...
Be aware of running abrt on RHEL 6.7 when using Satellite 6. abrt can get stuck in a loop of reporting crashes that itself causes. This generates enormous amounts of traffic to candlepin (tomcat) (postgresql) and foreman causing it to consume fantastic amounts of CPU and memory.
Check if you are affected on RHEL 6.7 by looking at number of crash counts:
# abrt-cli list|grep -i count
# for item in abrtd abrt-oops abrt-ccpp; do service $item stop; chkconfig $item off; done