XenServer 6.5 SP1 heavy IO traffic/ CPU utilization with open vswitch technology.
Ovs-vswitchd daemon running on the XS was showing high CPU usage and performance tab shows heavy CPU usage.
 

Citrix is not responsible for and does not endorse or accept any responsibility for the contents or your use of these third party Web sites. Citrix is providing these links to you only as a convenience, and the inclusion of any link does not imply endorsement by Citrix of the linked Web site. It is your responsibility to take precautions to ensure that whatever Web site you use is free of viruses or other harmful items.

Disabled NLP from Virtual connect manager. http://h20564.www2.hpe.com/hpsc/doc/public/display?docId=emr_na-c03358981&sp4ts.oid=3552695

---

Problem Cause

--- High number of flows causing the daemon to use up the CPU cycles. Look at the ovs-dpctl show o/p
 
 cat ovs-dpctl-show-s.out
system@ovs-system:
        lookups: hit:3094575866 missed:217618238 lost:0
        flows: 16384
 
What is causing this high flows and how to identify
 
Looking at the flows using ovs-appctl bridge/dump-flows bridgename( collected in bugreport as a file named ovs-appctl-dpif%dump-flows-xapi1.out), we can see a pattern in the ovs flows. 90 % of the flows are like below
 
skb_priority(0),in_port(5),eth(src=34:64:a9:e2:95:1d,dst=01:14:c2:44:1e:cc),eth_type(0x8100),vlan(vid=3907/0xfff,pcp=0/0x0,cfi=1/1),encap(), packets:4, bytes:320, used:2.080s, actions:drop
skb_priority(0),in_port(5),eth(src=34:64:a9:e2:95:1d,dst=01:14:c2:44:1e:cc),eth_type(0x8100),vlan(vid=3825/0xfff,pcp=0/0x0,cfi=1/1),encap(), packets:4, bytes:320, used:2.084s, actions:drop
skb_priority(0),in_port(2),eth(src=34:64:a9:e2:95:09,dst=01:14:c2:44:1e:cc),eth_type(0x8100),vlan(vid=2107/0xfff,pcp=0/0x0,cfi=1/1),encap(), packets:5, bytes:400, used:1.712s, actions:1
skb_priority(0),in_port(2),eth(src=34:64:a9:e2:95:09,dst=01:14:c2:44:1e:cc),eth_type(0x8100),vlan(vid=2662/0xfff,pcp=0/0x0,cfi=1/1),encap(), packets:19, bytes:1520, used:1.676s, actions:1
 
Looking at the pattern, we could identify this is related to Network Loop Protection (NLP). 01:14:c2:44:1e:cc is a reserved HP multicast address.

http://h20564.www2.hpe.com/hpsc/doc/public/display?docId=emr_na-c03358981&sp4ts.oid=3552695