Tracking anomaly propagation at the network level

1. A method comprising:

monitoring, by a device in a network and based on receiving registration information for a particular application, one or more application-centric metrics regarding network traffic between an application server hosting the particular application and a client device accessing the application server, the application-centric metrics indicative of a behavior of the particular application;

detecting, by the device, an application-centric anomaly based on the monitored one or more application-centric metrics by comparing the behavior of the particular application to a machine learning based model that models traffic behaviors of the particular application; and

causing, by the device, an anomaly mitigation action to be performed in the network, in response to detecting the application-centric anomaly by comparing the behavior of the particular application to the machine learning based model.

2. The method as in claim 1 , wherein the one or more application-centric metrics comprise at least one of: a number of unique connections associated with the particular application, a flow duration of a traffic flow associated with the particular application, or a volume of the network traffic associated with the particular application.

3. The method as in claim 1 , further comprising:

receiving, at the device, a request to monitor the one or more application-centric metrics regarding network traffic associated with the particular application from one of: the application server hosting the particular application or the client device that accesses the application server.

4. The method as in claim 3 , wherein the device receives the request from the application server or the client device via an application registration device in the network.

5. The method as in claim 1 , wherein causing the anomaly mitigation action to be performed in the network comprises:

blocking, by the device, new traffic associated with the particular application that was not previously observed by the device.

6. The method as in claim 1 , wherein causing the anomaly mitigation action to be performed in the network comprises:

requesting, by the device, that a process associated with the particular application stop sending the network traffic.

7. The method as in claim 1 , wherein causing the anomaly mitigation action to be performed in the network comprises:

requesting, by the device, that another device executing a process associated with the particular application shut down the process.

8. The method as in claim 1 , wherein causing the anomaly mitigation action to be performed in the network comprises:

providing, by the device, an indication of the detected application-centric anomaly to another device in the network, wherein the other device performs the anomaly mitigation action in response to receiving the indication.

9. The method as in claim 8 , wherein the mitigation action corresponds to the other device stopping execution of a process associated with the particular application.

10. A method, comprising:

registering, by a first device in a network, a particular application with a second device in the network;

specifying, by the first device, one or more application-centric metrics to be monitored by the second device that are indicative of a behavior of the particular application;

specifying, by the first device, one or more anomaly conditions for the application-centric metrics, wherein the second device uses the specified one or more anomaly conditions and the specified one or more application-centric metrics to detect an anomaly in network traffic between an application server hosting the particular application and a client device accessing the application server that is associated with the particular application by comparing the behavior of the particular application to a machine learning based model that models traffic behaviors of the particular application; and

receiving, at the first device, an indication of the detected anomaly from the second device.

11. The method as in claim 10 , wherein the one or more application-centric metrics comprise at least one of: a number of unique connections associated with the particular application, a flow duration of a traffic flow associated with the particular application, or a volume of the network traffic associated with the particular application.

12. The method as in claim 10 , wherein the first device registers the application with the second device via an application registration device in the network.

13. The method as in claim 10 , wherein receiving the indication of the detected anomaly from the second device comprises:

receiving, from the second device, a request to perform an anomaly mitigation action.

14. The method as in claim 13 , wherein the anomaly mitigation action corresponds to the first device shutting down the application.

15. The method as in claim 10 , further comprising:

performing, by the first device, an anomaly mitigation action, based on the detected anomaly and in response to receiving the indication of the detected anomaly.

16. The method as in claim 15 , wherein the anomaly mitigation action corresponds to the first device shutting down the application.

17. An apparatus, comprising:

one or more network interfaces to communicate with a network;

a processor coupled to the one or more network interfaces and configured to execute a process; and

a memory configured to store the process executable by the processor, the process when executed operable to:

monitor, based on receiving registration information for a particular application, one or more application-centric metrics regarding network traffic between an application server hosting the particular application and a client device accessing the application server, the application-centric metrics indicative of a behavior of the particular application;

detect an application-centric anomaly based on the monitored one or more application-centric metrics by comparing the behavior of the particular application to a machine learning based model that models traffic behaviors of the particular application; and

cause an anomaly mitigation action to be performed in the network, in response to detecting the application-centric anomaly by comparing the behavior of the particular application to the machine learning based model.

18. The apparatus as in claim 17 , wherein the one or more application-centric metrics comprise at least one of: a number of unique connections associated with the particular application, a flow duration of a traffic flow associated with the particular application, or a volume of the network traffic associated with the particular application.

19. The apparatus as in claim 17 , wherein the apparatus causes the anomaly mitigation action to be performed by at least one of: blocking new traffic associated with the particular application that was not previously observed by the apparatus, requesting that a process associated with the particular application stop sending the network traffic, or requesting that a device executing the process associated with the particular application shut down the process associated with the particular application.

20. The apparatus as in claim 17 , wherein the apparatus causes the anomaly mitigation action to be performed by:

providing an indication of the detected application-centric anomaly to another device in the network, wherein the other device performs the anomaly mitigation action in response to receiving the indication.

21. An apparatus, comprising:

one or more network interfaces to communicate with a network;

a processor coupled to the one or more network interfaces and configured to execute a process; and

a memory configured to store the process executable by the processor, the process when executed operable to:

register a particular application with a device in the network;

specify one or more application-centric metrics to be monitored by the device that are indicative of a behavior of the particular application;

specify one or more anomaly conditions for the application-centric metrics, wherein the device uses the specified one or more anomaly conditions and the specified one or more application-centric metrics to detect an anomaly in network traffic between an application server hosting the particular application and a client device accessing the application server that is associated with the particular application by comparing the behavior of the particular application to a machine learning based model that models traffic behaviors of the particular application; and

receive an indication of the detected anomaly from the second device.

22. The apparatus as in claim 21 , wherein the one or more application-centric metrics comprise at least one of: a number of unique connections associated with the particular application, a flow duration of a traffic flow associated with the particular application, or a volume of the network traffic associated with the particular application.

23. The apparatus as in claim 21 , wherein the process when executed is further operable to:

perform an anomaly mitigation action, based on the detected anomaly and in response to receiving the indication of the detected anomaly.

24. The apparatus as in claim 21 , wherein the apparatus receives the indication of the detected anomaly by:

receiving a request to perform an anomaly mitigation action from the device.