Endpoint malware detection using an event graph

1. A computer program product for detecting malware on an endpoint in an enterprise network, the computer program product comprising computer executable code embodied in a non-transitory computer readable medium that, when executing on the endpoint, performs the steps of:

instrumenting the endpoint to monitor a number of causal relationships among a number of computing objects at a plurality of logical locations within a computing environment related to the endpoint;

selecting a set of logical locations from the plurality of logical locations, the set of logical locations excluding at least one logical location of the plurality of logical locations associated with a known, good process;

recording a sequence of events causally relating the number of computing objects at the set of logical locations;

creating an event graph based on the sequence of events;

applying a malware detection rule to the event graph; and

remediating the endpoint when the malware detection rule and the event graph indicate a compromised security state.

2. The computer program product of claim 1 , wherein selecting the set of logical locations includes selecting a group of logical locations from the plurality of logical locations based on exposure to an external environment.

3. The computer program product of claim 1 , wherein selecting the set of logical locations includes selecting a group of logical locations from the plurality of logical locations based on reputation.

4. The computer program product of claim 1 , wherein selecting the set of logical locations includes increasing a number of logical locations in the set of logical locations based on an inconsistency between a reputation of a first process and a reputation of a second process calling the first process.

5. A method for malware detection comprising:

instrumenting a first endpoint to monitor a number of causal relationships among a number of computing objects at a first set of logical locations within a computing environment related to the first endpoint;

excluding, from the first set of logical locations, at least one logical location associated with a known, good process;

recording a sequence of events causally relating the number of computing objects at the first set of logical locations excluding the at least one logical location associated with the known, good process;

creating an event graph based on the sequence of events;

applying a malware detection rule to the event graph; and

remediating the first endpoint when the malware detection rule and the event graph indicate a compromised security state.

6. The method of claim 5 , wherein the first set of logical locations includes logical locations associated with processes of unknown reputation.

7. The method of claim 5 , further comprising adding one or more logical locations to the first set of logical locations in response to a detected increase in security risk.

8. The method of claim 5 , further comprising removing a logical location from the first set of logical locations in response to a detected decrease in security risk.

9. The method of claim 5 , further comprising filtering one or more of the events in the sequence of events according to reputation.

10. The method of claim 5 , wherein the first set of logical locations includes at least one endpoint separate from the first endpoint.

11. The method of claim 5 , wherein the first set of logical locations includes at least one programming interface to a human interface device.

12. The method of claim 5 , further comprising identifying one of the number of computing objects as a cause of the compromised security state and remediating the one of the number of computing objects.

13. The method of claim 5 , wherein the number of causal relationships include a data flow.

14. The method of claim 5 , wherein the number of causal relationships include a control flow.

15. The method of claim 5 , wherein the number of causal relationships include a network flow.

16. The method of claim 5 , wherein the one or more computing objects include one or more types of computing objects selected from a group consisting of a data file, a process, an application, a registry entry, a network address, and a peripheral device.

17. The method of claim 5 , wherein a number of events within the sequence of events are preserved for a predetermined time window, and further wherein the predetermined time window has a different duration for at least two different types of computing objects.

18. An endpoint comprising:

a network interface;

a memory; and

a processor configured by computer executable code stored in the memory to detect malware by performing the steps of instrumenting the endpoint to monitor a number of causal relationships among a number of computing objects at a set of logical locations within a computing environment related to the endpoint, excluding, from the set of logical locations, at least one logical location of the set of logical locations associated with a known, good process, recording a sequence of events causally relating the number of computing objects at the set of logical locations excluding the at least one logical location of the set of logical locations associated with the known, good process, creating an event graph based on the sequence of events, applying a malware detection rule to the event graph, and remediating the endpoint when the malware detection rule and the event graph indicate a compromised security state.

19. The endpoint of claim 18 , wherein the processor is further configured to adjust the set of logical locations by adding a new logical location, removing an existing logical location, or changing a level of filtering at one of the set of logical locations according to a security state of the endpoint.

20. The endpoint of claim 18 , wherein the number of causal relationships include one or more of a data flow, a control flow, and a network flow.