Function-targeted virtual machine switching

• US 9,304,795 B2
• Filed: 01/15/2013
• Issued: 04/05/2016
• Est. Priority Date: 01/15/2013
• Status: Expired due to Fees

• Current Assignee: Empire Technology Development LLC (Allied Inventors Management, LLC)
• Original Assignee: Empire Technology Development LLC (Allied Inventors Management, LLC)
• Inventors: Kruglick, Ezekiel
• Primary Examiner(s): Poltorak, Peter
• Assistant Examiner(s): Doan, Huan V
• Application Number:
  US13/977,670
• Publication Number:
  US 20140359778A1
• Time in Patent Office: 1,176 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 2009/45591 Monitoring or debugging support
• CPC Class Codes: G06F 21/577 Assessing vulnerabilities and evaluating computer system security
• CPC Class Codes: G06F 9/45533 Hypervisors; Virtual machine monitors
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

Technologies are provided for function-targeted virtual machine switching. In some examples, function usage times on a virtual machine (VM) may be profiled by a virtual machine manager (VMM) and used to manage VM switching in order to preferentially switch VMs during specific targeted functions. The targeted functions and/or VM switching preferences may be adjusted over time in order to provide switching unpredictability, for example to frustrate side-channel attackers by forcing the attackers to gather data for much longer periods of time (e.g., weeks or months) if they want to detect or attack.

Claims

1. A method for function-targeted virtual machine switching in a datacenter, the method comprising:

profiling a process executed on a first virtual machine that is being executed on a physical server;

building a probability table based on the profiling, wherein the probability table includes information regarding one or more functions that are associated with the executed process and a virtual machine switching probability, wherein the one or more functions are included in the process and are executed sequentially as part of the execution of the process;

selecting at least one function from the probability table, configured to execute on the first virtual machine;

adjusting the virtual machine switching probability associated with the at least one selected function using a slow alteration input corresponding to a desired rate of probability change, wherein the slow alteration input includes at least one aperiodic pseudorandom function having minimum derivatives that meet the desired rate of probability change;

in response to a detection that the at least one selected function is executing on the first virtual machine, switching from the first virtual machine to a second virtual machine that is being executed on a same physical server as the first virtual machine based on the adjusted virtual machine switching probability to prevent an attack; and

executing, on the second virtual machine, one or more functions configured to execute on the second virtual machine.

2. The method of claim 1, further comprising, switching from the first virtual machine to the second virtual machine based on a virtual machine switch window.

3. The method of claim 1, further comprising:

adjusting another virtual machine switching probability associated with another function configured to execute on the first virtual machine; and

when the other function is executing on the first virtual machine, switching from the first virtual machine to another virtual machine based on the adjusted other virtual machine switching probability.

4. The method of claim 1, wherein the desired rate of probability change is based on a time for obtaining statistically-sufficient data for a side-channel attack.

5. The method of claim 1, further comprising selecting the at least one function based on a vulnerability table that lists functions that are vulnerable or resistant to an attack.

6. The method of claim 5, further comprising building the vulnerability table by monitoring for at least one of functions with specific formats and functions executing on the first virtual machine when high-value data is processed.

7. A virtual machine manager (VMM) for function-targeted virtual machine switching in a datacenter, the VMM comprising:

a profiling module configured to profile a process executed on a first virtual machine that is being executed on a physical server; and

a processing module configured to;

build a probability table based on the profiling, wherein the probability table includes information regarding one or more functions that are associated with the executed process and a virtual machine switching probability, wherein the one or more functions are included in the process and are executed sequentially as part of the execution of the process;

select at least one function from the probability table;

adjust the virtual machine switching probability associated with the at least one selected function using a slow alteration input corresponding to a desired rate of probability change, wherein the slow alteration input includes at least one aperiodic pseudorandom function having minimum derivatives that meet the desired rate of probability change;

in response to a detection that the at least one selected function is executing on the first virtual machine, switch from the first virtual machine to a second virtual machine that is being executed on a same physical server as the first virtual machine based on the adjusted virtual machine switching probability to prevent an attack; and

execute, on the second virtual machine, one or more functions configured to execute on the second virtual machine.

8. The VMM of claim 7, wherein the desired rate of probability change is based on a time for obtaining statistically-sufficient data for a side-channel attack.

9. The VMM of claim 7, wherein the first virtual machine and the second virtual machine execute on different physical servers.

10. The VMM of claim 9, wherein the first virtual machine and the second virtual machine execute at different datacenters.

11. A cloud-based datacenter configured to employ function-targeted virtual machine switching, the datacenter comprising:

a plurality of virtual machines; and

a physical server configured to execute at least a first virtual machine and a second virtual machine in the plurality of virtual machines, the physical server comprising;

a memory; and

a processor coupled to the memory, wherein the processor comprises;

a profiling module configured to;

profile a process executed on the first virtual machine; and

build a probability table based on the profiling, wherein the probability table includes information regarding one or more functions that are associated with the executed process and a virtual machine switching probability, wherein the one or more functions are included in the process and are executed sequentially as part of the execution of the process; and

a function select module configured to;

select at least one function from the probability table, configured to execute on the first virtual machine;

adjust the virtual machine switching probability associated with the at least one selected function using a slow alteration input corresponding to a desired rate of probability change, wherein the slow alteration input includes at least one aperiodic pseudorandom function having minimum derivatives that meet the desired rate of probability change;

in response to a detection that the at least one selected function is executing on the first virtual machine, switch from the first virtual machine to the second virtual machine based on the adjusted virtual machine switching probability to prevent an attack; and

execute, on the second virtual machine, one or more functions configured to execute on the second virtual machine.

12. The datacenter of claim 11, wherein the function select module is further configured to switch from the first virtual machine to the second virtual machine based on a virtual machine switch window.

13. The datacenter of claim 11, wherein the function select module is further configured to:

adjust another virtual machine switching probability associated with another function configured to execute on the first virtual machine; and

in response to a detection that the other function is executing on the first virtual machine, switch from the first virtual machine to another virtual machine based on the adjusted other virtual machine switching probability.

14. The datacenter of claim 11, wherein the desired rate of probability change is based on a time required to obtain statistically-sufficient data for a side-channel attack.

15. The datacenter of claim 11, wherein the function select module is further configured to select the at least one function based on a vulnerability table.

16. The datacenter of claim 15, wherein the function select module is further configured to build the vulnerability table by monitoring for at least one of functions with specific formats and functions executing on the first virtual machine when high-value data is processed.