Distributed denial-of-service attack mitigation with reduced latency

1. An apparatus for mitigating a distributed denial-of-service (DDoS) attack in a networked computing system, the apparatus comprising:

at least one detector in operative communication with a corresponding router in the networked computing system, the at least one detector being configured:

to obtain network traffic flow information from the router regarding current data traffic to at least one host;

to employ a machine learning mechanism to inspect one or more traffic patterns prior to the DDoS attack and develop a prediction model to estimate a likelihood of a future DDoS attack;

to compare the current data traffic to at least one of the one or more stored traffic patterns associated with at least one prior DDoS attack in response to a prediction of the DDoS attack, each traffic pattern characterizing network traffic flow over a predefined period of time; and

to generate an output indicative of a match between the current data traffic and at least one of the stored traffic patterns; and

at least one mitigation unit in operative communication with the at least one detector, the at least one mitigation unit being configured: to receive the output indicative of the match between the current data traffic and at least one of the stored traffic patterns; and to initiate at least one DDoS attack mitigation action in response to the received output prior to detection of an actual DDoS attack to the at least one host.

2. The apparatus of claim 1 , wherein the at least one mitigation unit is further configured: to monitor the current data traffic for at least a prescribed period of time after initiation of the at least one DDoS attack mitigation action; to determine whether an actual DDoS attack to the at least one host has occurred within the prescribed period of time; to continue implementing the at least one DDoS attack mitigation action when it is determined that a DDoS attack has occurred within the prescribed period of time; and to disable the at least one DDoS attack mitigation action when it is determined that a DDoS attack to the at least one host has not occurred within the prescribed period of time.

3. The apparatus of claim 2 , wherein the at least one mitigation unit is further configured to record that a DDoS attack mitigation action was needed when it is determined that a DDoS attack has occurred within the prescribed period of time.

4. The apparatus of claim 2 , wherein the at least one mitigation unit is further configured to record that a DDoS attack mitigation action was not needed when it is determined that a DDoS attack has not occurred within the prescribed period of time.

5. The apparatus of claim 1 , wherein the at least one detector is further configured: to monitor network information from the router regarding data traffic from a prior networking session; to determine that a DDoS attack has occurred to at least one victim Internet Protocol (IP) address; to obtain from the router network traffic flow records to the victim IP address for a prescribed amount of time prior to the DDoS attack; to store pre-DDoS attack traffic patterns.

6. The apparatus of claim 5 , wherein the at least one detector is further configured to generate the prediction model for estimating the likelihood of the future DDoS attack as a function of the stored pre-DDoS attack traffic patterns, the output generated by the at least one detector being based at least in part on the prediction model.

7. The apparatus of claim 1 , wherein the at least one mitigation unit is further configured: to monitor the current data traffic for at least a prescribed period of time after initiation of the at least one DDoS attack mitigation action; to record that a DDoS attack mitigation action was not needed when it is determined that a DDoS attack has not occurred within the prescribed period of time and to record that a DDoS attack mitigation action was needed when it is determined that an actual DDoS attack has occurred within the prescribed period of time; and wherein the at least one detector is configured to update the prediction model based on recorded information regarding whether or not the DDoS attack mitigation was needed to thereby enhance an accuracy of the prediction model.

8. The apparatus of claim 1 , wherein the traffic patterns associated with at least one prior DDoS attack are stored in a database in operative communication with the at least one detector.

9. The apparatus of claim 1 , wherein the at least one DDoS attack mitigation action initiated by the at least one mitigation unit comprises at least one of rate-limiting the current data traffic, discarding packets from the current data traffic, proxying a connection from a sender to a victim, presenting a challenge to the sender, and performing deep packet inspection on at least a subset of packets in the current data traffic constituting malicious traffic flow.

10. The apparatus of claim 1 , the apparatus further comprising short-term NetFlow storage, the at least one detector being configured to obtain one or more traffic flow records to a victim IP address for at least one minute prior to the DDoS attack, wherein the one or more traffic flow records are obtained from the short-term NetFlow storage.

11. A method for mitigating a distributed denial-of-service (DDoS) attack in a networked computing system, the method comprising:

receiving network traffic flow information from a router in the networked computing system regarding current data traffic to at least one host;

employing a machine learning mechanism to inspect one or more traffic patterns prior to the DDoS attack and develop a prediction model to estimate a likelihood of a future DDoS attack;

comparing the current data traffic to at least one of the one or more stored traffic patterns associated with at least one prior DDoS attack in response to a prediction of the DDoS attack, each traffic pattern characterizing network traffic flow over a predefined period of time;

generating an output signal indicative of a match between the current data traffic and at least one of the stored traffic patterns; and

initiating at least one DDoS attack mitigation action in response to the received output signal prior to detection of an actual DDoS attack to the at least one host.

12. The method of claim 11 , further comprising:

monitoring the current data traffic for at least a prescribed period of time after initiation of the at least one DDoS attack mitigation action;

determining whether an actual DDoS attack to the at least one host has occurred within the prescribed period of time;

continuing to implement the at least one DDoS attack mitigation action when it is determined that a DDoS attack has occurred within the prescribed period of time; and

disabling the at least one DDoS attack mitigation action when it is determined that a DDoS attack to the at least one host has not occurred within the prescribed period of time.

13. The method of claim 12 , further comprising recording that a DDoS attack mitigation action was needed when it is determined that a DDoS attack has occurred within the prescribed period of time.

14. The method of claim 12 , further comprising recording that a DDoS attack mitigation action was not needed when it is determined that a DDoS attack has not occurred within the prescribed period of time.

15. The method of claim 11 , further comprising:

monitoring network information from the router regarding data traffic from a prior networking session;

determining whether a DDoS attack has occurred to at least one victim Internet Protocol (IP) address;

when it is determined that a DDoS attack has occurred to the at least one victim IP address, receiving from the router network traffic flow records to the victim IP address for a prescribed amount of time prior to the DDoS attack; and

storing pre-DDoS attack traffic patterns.

16. The method of claim 15 , further comprising generating the prediction model for estimating the likelihood of the future DDoS attack as a function of the stored pre-DDoS attack traffic patterns, the output signal being generated as a function of the prediction model.

17. The method of claim 11 , further comprising:

monitoring the current data traffic for at least a prescribed period of time after initiation of the at least one DDoS attack mitigation action;

recording that a DDoS attack mitigation action was not needed when it is determined that a DDoS attack has not occurred within the prescribed period of time;

recording that a DDoS attack mitigation action was needed when it is determined that an actual DDoS attack has occurred within the prescribed period of time; and

updating the prediction model based on recorded information regarding whether or not the DDoS attack mitigation was needed to thereby enhance an accuracy of the prediction model.

18. The method of claim 11 , wherein initiating the at least one DDoS attack mitigation action comprises at least one of rate-limiting the current data traffic, discarding packets from the current data traffic, proxying a connection from a sender to a victim, presenting a challenge to the sender, and performing deep packet inspection on at least a subset of packets in the current data traffic constituting malicious traffic flow.

19. The method of claim 11 , the method further comprising obtaining one or more traffic flow records to a victim IP address for at least one minute prior to the DDoS attack, wherein the one or more traffic flow records are obtained from short-term NetFlow storage.

20. A non-transitory computer readable medium comprising computer executable instructions which when executed by a computer cause the computer to perform a method of:

receiving network traffic flow information from a router in a networked computing system regarding current data traffic to at least one host;

employing a machine learning mechanism to inspect one or more traffic patterns prior to the DDoS attack and develop a prediction model to estimate a likelihood of a future DDoS attack;

comparing the current data traffic to at least one of the one or more stored traffic patterns associated with at least one prior DDoS attack in response to a prediction of the DDoS attack, each traffic pattern characterizing network traffic flow over a predefined period of time;

generating an output signal indicative of a match between the current data traffic and at least one of the stored traffic patterns; and

initiating at least one DDoS attack mitigation action in response to the received output signal.