I am running nginx process on rhel/CentOS release 6.5 (Final). I can see ksoftirqd and migration processes are taking much of CPU due to which load on my server increases by great extent. I have made irqbalance service off. In cat /proc/interrupts output i can see out of 60 CPUs only 2 CPUs and their respective cores are getting used.Is this the reason for causing high load on server whenever there are more number of interrupts/socket connections on server. Can someone please help me to manage these interrupts so that at peak time there is less cpu load due increased number of interrupts. I have heard of taskset but it works only on pids. Other than upgrading to rhel7 / centos 6.8 can someone please suggest workaroung to handle these inturrupts in better way.