Virtual machine snapshotting and damage containment
Some embodiments provide a system that manages the execution of a virtual machine. During operation, the system takes a series of snapshots of the virtual machine during execution of the virtual machine. If an abnormal operation of the virtual machine is detected, the system spawns a set of snapshot instances from one of the series of snapshots, wherein each of the snapshot instances is executed with one of a set of limitations. Next, the system determines a source of the abnormal operation using a snapshot instance from the snapshot instances that does not exhibit the abnormal operation. Finally, the system updates a state of the virtual machine using the snapshot instance.
1. A computer-implemented method for managing the execution of a virtual machine, comprising:
taking a series of snapshots of the virtual machine during execution of the virtual machine, wherein each snapshot captures a state of the virtual machine at a corresponding time;
if an abnormal operation of the virtual machine is detected:
spawning a set of snapshot instances from one of the series of snapshots, wherein spawning the set of snapshot instances comprises, for each snapshot instance in the set of snapshot instances, concurrently executing the snapshot instance using a different set of limitations, wherein each limitation in the set blocks a different potential source of the abnormal operation so that the snapshot using the limitation does not process incoming events from the potential source;
determining a source of the abnormal operation using a snapshot instance from the snapshot instances that does not exhibit the abnormal operation; and
updating a state of the virtual machine using the snapshot instance.
2. The computer-implemented method of claim 1 , wherein updating the state of the virtual machine using the snapshot instance involves:
accelerating the execution of the snapshot instance to reach a current state of the virtual machine; and
replacing the virtual machine with the snapshot instance.
3. The computer-implemented method of claim 2 , wherein accelerating the execution of the snapshot instance involves:
caching incoming events associated with the virtual machine; and
executing the snapshot instance at an accelerated rate using the cached incoming events.
4. The computer-implemented method of claim 1 , further comprising:
selecting a snapshot from the series of snapshots to execute as the set of snapshot instances; and
if the source of the abnormal operation cannot be determined from the snapshot, selecting an earlier snapshot from the series of snapshots to execute as the set of snapshot instances.
5. The computer-implemented method of claim 1 , wherein the snapshots are taken at periodic intervals during execution of the virtual machine.
6. The computer-implemented method of claim 1 , wherein the potential source of the abnormal operation corresponds to at least one of a port and an address.
7. The method of claim 1 , wherein a limitation corresponds to at least one of a port, address, disk input, keyboard input, and a remote shell.
8. The method of claim 1 , wherein a limitation corresponds to at least one of a disk input, keyboard input, and a remote shell.
9. The method of claim 1 , wherein a given set of the limitations corresponds to a port, address, disk input, keyboard input, and a remote shell.
10. A system for managing the execution of a virtual machine, comprising:
a computing system that performs operations for:
an introspection module configured to monitor an execution of the virtual machine for abnormal operation; and
a domain manager configured to take a series of snapshots of the virtual machine during the execution of the virtual machine, wherein each snapshot captures a state of the virtual machine at a corresponding time,
wherein if an abnormal operation of the virtual machine is detected:
the domain manager is further configured to spawn a set of snapshot instances from one of the series of snapshots, wherein spawning the set of snapshot instances comprises, for each snapshot instance in the set of snapshot instances, concurrently executing the snapshot instance using a different set of limitations, wherein each limitation in the set blocks a different potential source of the abnormal operation so that the snapshot using the limitation does not process incoming events from the potential source; and
the introspection module is further configured to determine a source of the abnormal operation using a snapshot instance from the snapshot instances that does not exhibit the abnormal operation.
11. The system of claim 10 , wherein the domain manager is further configured to update a state of the virtual machine using the snapshot instance.
12. The system of claim 11 , wherein updating the state of the virtual machine using the snapshot instance involves:
accelerating the execution of the snapshot instance to reach a current state of the virtual machine; and
replacing the virtual machine with the snapshot instance.
13. The system of claim 12 , wherein accelerating the execution of the snapshot instance involves:
caching incoming events associated with the virtual machine; and
executing the snapshot instance at an accelerated rate using the cached incoming events.
14. The system of claim 10 , wherein the domain manager is further configured to:
select a snapshot from the series of snapshots to execute as the set of snapshot instances; and
if the source of the abnormal operation cannot be determined from the snapshot, select an earlier snapshot from the series of snapshots to execute as the set of snapshot instances.
15. The system of claim 10 , wherein the snapshots are taken at periodic intervals during execution of the virtual machine.
16. The system of claim 10 , wherein the potential source of the abnormal operation corresponds to at least one of a port and an address.
17. A non-transitory computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for managing the execution of a virtual machine, the method comprising:
taking a series of snapshots of the virtual machine during execution of the virtual machine, wherein each snapshot captures a state of the virtual machine at a corresponding time;
if an abnormal operation of the virtual machine is detected:
spawning a set of snapshot instances from one of the series of snapshots, wherein spawning the set of snapshot instances comprises, for each snapshot instance in the set of snapshot instances, concurrently executing the snapshot instance using a different set of limitations, wherein each limitation in the set blocks a different potential source of the abnormal operation so that the snapshot using the limitation does not process incoming events from the potential source;
determining a source of the abnormal operation using a snapshot instance from the snapshot instances that does not exhibit the abnormal operation; and
updating a state of the virtual machine using the snapshot instance.
18. The computer-readable storage medium of claim 17 , wherein updating the state of the virtual machine using the snapshot instance involves:
accelerating the execution of the snapshot instance to reach a current state of the virtual machine; and
replacing the virtual machine with the snapshot instance.
19. The computer-readable storage medium of claim 18 , wherein accelerating the execution of the snapshot instance involves:
caching incoming events associated with the virtual machine; and
executing the snapshot instance at an accelerated rate using the cached incoming events.
20. The computer-readable storage medium of claim 17 , the method further comprising:
selecting a snapshot from the series of snapshots to execute as the set of snapshot instances; and
if the source of the abnormal operation cannot be determined from the snapshot, selecting an earlier snapshot from the series of snapshots to execute as the set of snapshot instances.