System and method for starting a cloud-based virtualization system with hypervisor and virtual machine monitor

1. A computing cloud comprising:

a plurality of computing nodes connected into a cluster, each node having a Hypervisor with the highest privilege level, and a Virtual Machine Monitor (VMM) running with no higher privilege level than the Hypervisor, and Virtual Machines (VMs) running on a non-privileged level;

each Virtual Machine having a Guest Operating System running within it;

each node having a host operating system (HOS), each HOS having access to at least some hardware devices of the corresponding computing node under supervision of the Hypervisor; and

the VMs of the corresponding node using the HOS of its node to access the hardware devices,

wherein the Hypervisor processes all interrupts and exceptions and dispatches them to the HOS and VMMs, and

wherein the Hypervisor is responsible for interrupt forwarding, for context switching between the HOS and the VMMs, and for resource scheduling.

2. The computing cloud of claim 1 , wherein the Hypervisor reloads an Interrupt Descriptor Table (IDT) and protects it from modification by the HOS and the VMMs.

3. The computing cloud of claim 2 , wherein the Hypervisor captures an IDT pointer and replaces it with its own.

4. The computing cloud of claim 2 , wherein the Hypervisor performs any of:

allocation and freeing pages for the IDT;

modification of IDT entries;

emulation of IDT writes;

protection of the IDT and emulation of direct writes to the IDT;

passthrough device emulation without interaction with the HOS;

emulation and maintenance of an interrupt subscription list for HOS components.

5. The computing cloud of claim 1 , wherein the Hypervisor provides service hypercalls for any of:

subscription for a particular interrupt;

context switch from the HOS into a VMM;

context switch from a VMM into the HOS;

context switch from one VMM to another VMM;

memory management between the VMMs;

time scheduling between the VMMs;

creating a new VMM and a new VM;

deleting a VMM and a VM;

subscription for a particular I/O device; and

I/O operation with the particular I/O device.

6. The computing cloud of claim 1 , wherein the HOS is migratable from one node to another node in online or offline mode.

7. The computing cloud of claim 1 , wherein the VMs are migratable from one node to another node.

8. The computing cloud of claim 1 , wherein HOS or its VM can migrate on-the-fly after system failure or overload.

9. The computing cloud of claim 1 , wherein secure isolation enables any VM to communicate with the HOS via intermediation dedicated code rerouting messages between processes so that VM processes are not aware of HOS processes.

10. The computing cloud of claim 1 , wherein a HOS that becomes unsecure is terminated and a new HOS is launched in the computing cluster.

11. The computing cloud of claim 1 , wherein the hardware device is any of a mass storage device, a hardware interface device, a video device subsystem and a human interface device.

12. The computing cloud of claim 1 , wherein the cloud is a server farm.

13. The computing cloud of claim 1 , wherein an application memory area of at least one VM is protected from access by the hardware resources; and

the applications are trusted applications that cannot be observed or modified by other software of the computing cloud.

14. The computing cloud of claim 1 , further comprising a secure boot manager for starting a secure VM using trusted hypervisor.

15. The computing cloud of claim 1 , further comprising means for migrating the HOS from a node with a trusted hypervisor to another node with trusted hypervisor.

16. The computing cloud of claim 1 , further comprising at least one node with a trusted component.

17. The computing cloud of claim 16 , wherein the trusted component is a driver or the hypervisor.

18. A method of operating a computing cloud comprising:

initializing a plurality of computing nodes and connecting them into a cluster, each node having a Hypervisor with the highest privilege level, and a Virtual Machine Monitor (VMM) running with no higher privilege level than the Hypervisor, and Virtual Machines (VMs) running on a non-privileged level,

with each node having a host operating system (HOS), each HOS having access to at least some hardware devices of the corresponding computing node under supervision of the Hypervisor instantiating a Guest Operating System in each Virtual Machine; and

for the VMs of the corresponding node, using the HOS of its node to access the hardware devices,

wherein the Hypervisor processes all interrupts and exceptions and dispatches them to the HOS and VMM, and

wherein the Hypervisor is responsible for interrupt forwarding, context switching between the HOS and the VMMs, and for resource scheduling.

19. The method of claim 18 , further comprising migrating the HOS from one node to another node, the migrating comprising:

stopping HOS execution;

saving HOS CPU context;

saving contents of HOS memory;

loading saved HOS CPU context into a virtual CPU;

loading saved HOS memory into a VM;

resuming HOS execution.

20. The method of claim 19 , further comprising updating a configuration the HOS to conform to hardware changes after migration.

21. The method of claim 18 , further comprising migrating the HOS from one node to another node, the migrating comprising:

installing a boot manager component of the Hypervisor

restarting the HOS;

during host boot, having the Hypervisor take control before the HOS;

starting the HOS booting process under control of Hypervisor; and

resuming HOS execution.

22. The method of claim 21 , further comprising updating a configuration the HOS to conform to hardware changes after migration.

23. The method of claim 19 , further comprising migrating the HOS from one node to another node, the migrating comprising:

installing a boot manager component of the Hypervisor;

restarting the HOS;

during host boot, having the Hypervisor take control before the HOS;

initiating HOS installation process; and

resuming HOS execution.

24. The method of claim 23 , further comprising updating a configuration the HOS to conform to hardware changes after migration.