System and method for container assessment using sandboxing

1. A method for a software container, comprising:

performing static image scanning of a container image outside of a sandbox of a computing device;

assigning, based on the static scanning, a respective container image risk score to each static scanning event of the container image that is potentially malicious, each container image risk score indicating a likelihood that a corresponding static scanning event is malicious;

instantiating in the sandbox of the computing device: an operating system, a Berkeley Packet Filter (BPF) virtual machine within a kernel of the operating system, and a software container, wherein the software container is instantiated from the container image;

monitoring runtime behavior events of the software container, the monitoring performed by the kernel and at least partially performed by the BPF virtual machine;

assigning, based on the monitoring, a respective runtime risk score to each of the runtime behavior events that is potentially malicious, each runtime risk score indicating a likelihood that a corresponding behavior event is malicious; and

assigning an overall risk score to the software container that indicates a likelihood that the software container is malicious, the overall risk score corresponding to a highest severity risk score of the container image and runtime risk scores.

2. The method of claim 1 , wherein the software container is instantiated by a container engine, the method comprising:

further monitoring runtime behavior events of the software container from outside the sandbox by reviewing a runtime behavior log file from the container engine;

wherein said assigning the overall risk score is further based on said reviewing the runtime behavior log file.

3. The method of claim 1 , wherein said monitoring comprises:

monitoring at least one first type of behavior of the software container by the BPF virtual machine; and

monitoring a second type of behavior of the container that is different from the at least one first type of behavior.

4. The method of claim 3 , wherein the kernel is a LINUX kernel and the at least one first type of behavior includes one or more of:

system calls to the LINUX kernel;

requests to utilize LINUX security hooks; and

file system requests to the LINUX kernel.

5. The method of claim 4 , wherein the second type of behavior includes network traffic behavior associated with the software container.

6. The method of claim 5 , wherein said monitoring the second type of behavior of the container is performed by the kernel but outside of the BPF virtual machine.

7. The method of claim 5 , wherein said monitoring the second type of behavior of the container is performed by the BPF virtual machine.

8. The method of claim 1 , wherein the assigning of the container image and runtime risk scores and the overall risk score are performed outside of the sandbox.

9. A computing device comprising:

memory; and

a processor operatively connected to the memory, the processor configured to:

perform static image scanning of a container image outside of a sandbox of a computing device;

assign, based on the static scanning, a respective container image risk score to each static scanning event of the container image that is potentially malicious, each container image risk score indicating a likelihood that a corresponding static scanning event is malicious;

instantiate in the sandbox of the computing device: an operating system, a Berkeley Packet Filter (BPF) virtual machine within a kernel of the operating system, and a software container, the sandbox corresponding to a virtual machine that is different from the BPF virtual machine, wherein the software container is instantiated from the container image;

monitor runtime behavior events of the software container, the monitoring performed by the kernel and at least partially performed by the BPF virtual machine;

assign, based on the monitoring, a respective runtime risk score to each of the runtime behavior events that is potentially malicious, each runtime risk score indicating a likelihood that a corresponding behavior event is malicious; and

assign an overall risk score to the software container that indicates a likelihood that the software container is malicious, the overall risk score corresponding to a highest severity risk score of the container image and runtime risk scores.

10. The computing device of claim 9 , wherein the software container is instantiated by a container engine and the processor is configured to:

further monitor runtime behavior events of the software container from outside the sandbox by reviewing a runtime behavior log file from the container engine; and

further base the assignment of the overall risk score on the reviewing of the log file.

11. The computing device of claim 9 , wherein to monitor the runtime behavior events, the processor is configured to:

monitor at least one first type of behavior of the software container by the BPF virtual machine; and

monitor a second type of behavior of the container that is different from the at least one first type of behavior.

12. The computing device of claim 11 , wherein the kernel is a LINUX kernel and the at least one first type of behavior includes one or more of:

system calls to the LINUX kernel;

requests to utilize LINUX security hooks; and

file system requests to the LINUX kernel.

13. The computing device of claim 12 , wherein the second type of behavior includes network traffic behavior associated with the software container.

14. The computing device of claim 13 , wherein the monitoring of the second type of behavior of the container is performed by the kernel but outside of the BPF virtual machine.

15. The computing device of claim 13 , wherein the monitoring of the second type of behavior of the container is performed by BPF virtual machine.

16. The computing device of claim 9 , wherein the processor is configured to assign the container image and runtime risk scores and the overall risk score outside of the sandbox.