...

After trying out several things, finaly we can say that the main problem are the usage of SWAP and Overcommitting of Memory, probably the fair I/O scheduler in the Linux Kernel. After disabling Swap and overcommitting memory and enabling the noop I/O scheduler, no crash happened within 8 hours of above fio load.

What is the fix? 

The next Hypervisor release 1.03 will remove the need to upgrade the Kernel, Xen and so forth, as all these Versions don't differ in the crash behaviour. It will disable swap, overcommitting and enable the noop I/O scheduler. These steps should finaly stabilize the hypervisor and make it also ready for kernel upgrades. 

2017-01-10 Current Hypervisor State is BETA

...