HV seems to be under heavy load. Particularly for the KVM HVs, the STORAGENODE* process is using too much CPU resources.
STORAGENODE VMs are the actual storage controllers for Integrated Storage. Their usage of CPU and other resources is so that the VMs using Integrated Storage can access their disks, so that backups can be taken.
Typically, if a STORAGENODE VM has a high CPU usage, that would mean that some VMs have heavy disk access. These VMs would be using node(s) that are controllers by the STORAGENODE VM, and may not be on same HV or even in the same data store.
The STORAGENODE can be telnet into, and it runs a stripped-down diskless OS. However, simple commands like 'ps' and 'top' are available. 'top' can be used to see what disk devices may be using the most resources at the moment, and these can be looked up in the UI and, thus, traced back to the VM.
It should be easier and safer to check disk usage statistics in the OnApp UI first and determine what VMs listed may be loading down a particular STORAGENODE. Then, appropriate action could be determined afterward.