Detecting compute resource anomalies in a group of computing resources

1. A method for anomaly detection in a service provider environment, comprising:

determining a group of compute resources that share a set of attributes, wherein the compute resources include computing instances each comprising service agents, and the service agents provide attribute data regarding the computing instances that is used to group the compute resources, wherein the determining comprises:

weighting attributes of a set of attributes based on a relevancy of attributes in determining a grouping, wherein the weighting prioritizes a first attribute above a second attribute of the set of attributes based on a relevancy of the first attribute in determining the grouping; and

determining the group of compute resources that share the set of attributes based in part on the weighted attributes;

collecting initial attribute values of a plurality of attributes for compute resources in the group;

collecting subsequent attribute values of the plurality of attribute for compute resources in the group;

determining a difference between the plurality of the initial attribute values and the subsequent attribute values of the plurality of attributes of compute resources in the group to identify an anomaly in one of the computing instances; and

sending an anomaly detection event notification regarding the one of the computing instances to a client device associated with an administrator.

2. The method of claim 1 , wherein the group of compute resources that share the set of attributes further comprises: a plurality of compute resources associated with a load balancer, a plurality of automatically scaled compute resources, a plurality of compute resources activated from a common machine image, or a plurality of compute resources activated from a common configuration management operation.

3. The method of claim 1 , wherein the weighting of the attributes is based in part on a social distance.

4. The method of claim 1 , wherein the plurality of attributes include one or more parameters selected from configuration parameters, event parameters, and execution parameters.

5. The method of claim 1 , wherein the plurality of attributes are received from at least one of a network accessible Application Programming Interface (API) and an agent executing on a customer account.

6. A method, comprising:

identifying a plurality of compute resources in a service provider environment, the compute resources including computing instances that are susceptible to compute resource anomalies, using metadata from the service provider environment, wherein the metadata is provided from service agents associated with each of the compute resources and the metadata is used to identify attribute information for attributes of the computing instances;

weighting attributes of a set of attributes based on relevancy of the attributes;

determining a group of compute resources from the plurality of compute resources, wherein the group of compute resources is determined based at least in part by prioritizing weighted attributes in the weighted set of attributes;

determining a difference in one or more of a plurality of attributes of the compute resources of the group to identify an anomaly in one of the computing instances; and

outputting an anomaly detection notification containing the anomaly regarding the one of the computing instances to a client device associated with an administrator.

7. The method of claim 6 , wherein determining the group of compute resources comprises:

identifying the set of attributes for determining a grouping of compute resources with a predetermined level of confidence; and

selecting the group of similar compute resources from the plurality of compute resources based on the set of attributes.

8. The method of claim 7 , wherein the weighting of the attributes is based in part on a social distance.

9. The method of claim 6 , wherein determining the difference in one or more of the plurality of attributes of the compute resources of the group comprises:

determining a difference in one or more attributes of a compute resource or computing instance and one or more additional compute resources or computing instances in the group by comparing a corresponding attribute between an instance of the compute resource and instances of the additional compute resources.

10. The method of claim 6 , wherein determining the difference in one or more of the plurality of attributes of the compute resources of the group comprises:

determining a difference between initial values and current values of one or more of the plurality of attributes of the compute resources of the group, wherein the initial values were recorded in a time period preceding capture of the current values.

11. The method of claim 6 , wherein determining the difference in one or more of the plurality of attributes of the compute resources of the group comprises:

determining a difference between expected values and current values of one or more of the plurality of attributes of the compute resources of the group to identify an anomaly in the current values.

12. The method of claim 6 , wherein outputting the anomaly detection notification comprises:

generating a report that lists each computing resource determined to have a difference in one or more of the plurality of attributes and the value of the differing one or more attributes.

13. The method of claim 6 , wherein outputting the anomaly detection notification comprises:

triggering an intrusion detection system process.

14. The method of claim 6 , wherein outputting the anomaly detection notification comprises:

outputting the anomaly detection notification when the difference between one or more of the plurality of attributes of the compute resources of the group exceeds a predetermined threshold.

15. The method of claim 6 , wherein the plurality of attributes include one or more parameters selected from configuration parameters, event parameters, execution parameters and execution patterns.

16. The method of claim 6 , wherein the group of similar compute resources includes: one or more host computing instances of an automatic scaling group, one or more host computing instances from a machine image, or one or more host computing instances from a configuration management operation.

17. The method of claim 6 , wherein determining the difference in one or more of the plurality of attributes of the compute resources of the group comprises:

collecting baseline attribute values of the plurality of attributes for each compute resource in the group;

receiving an anomaly detection trigger event;

collecting updated attribute values of the plurality of attributes for the compute resources in the group; and

determining for the compute resources of the group whether at least one updated attribute value differs from a corresponding baseline attribute value of a respective one of the plurality of attributes.

18. A method for anomaly detection in a service provider environment, comprising:

determining a group of compute resources that share a set of attributes, wherein the compute resources include computing instances each comprising service agents, and the service agents provide attribute data regarding the computing instances that is used to group the compute resources, wherein the determining comprises:

weighting attributes of a set of attributes based on a relevancy of attributes in determining a grouping, wherein the weighting prioritizes a first attribute above a second attribute of the set of attributes based on a relevancy of the first attribute in determining the grouping; and

determining the group of compute resources that share the set of attributes based in part on the weighted attributes;

collecting initial attribute values of a plurality of attributes for compute resources in the group;

collecting subsequent attribute values of the plurality of attribute for compute resources in the group;

determining a difference between the plurality of the initial attribute values and the subsequent attribute values of the plurality of attributes of compute resources in the group;

detecting an anomaly based on the difference to identify an anomaly in one of the computing instances; and

triggering a remediation action based in part on detecting the anomaly, in order to mitigate the anomaly in the one of the computing instances.

19. The method of claim 18 , wherein the remediation action is triggered once the difference exceeds a predetermined threshold.

20. The method of claim 18 , wherein the group of compute resources that share the set of attributes further comprises: a plurality of compute resources associated with a load balancer, a plurality of automatically scaled compute resources, a plurality of compute resources activated from a common machine image, or a plurality of compute resources activated from a common configuration management operation.