Distributed denial of service (DDOS) based accelerated solution
Apparatuses, systems, and techniques for detecting that a host device is subject to a distributed denial of service (DDOS) attack using a machine learning (ML) detection system are described. A computing system includes a switch with port interfaces, a central processing unit (CPU) that implements a machine learning (ML) detection system, and network monitoring logic. The network monitoring logic can extract features from network data and send the extracted features to the ML detection system. The ML detection system determines whether the host device is subject to a DDOS attack using the extracted features. The ML detection system can send an alert to the host device responsive to a determination that the host device is subject to the DDOS attack.
1 . A switch comprising:
an acceleration hardware engine;
a plurality of port interfaces;
memory to store instructions and a configuration file that specifies one or more filters that extract for inclusion or remove from inclusion specified types of data; and
a central processing unit (CPU) operatively coupled to the memory and the plurality of port interfaces, wherein the CPU is to execute the instructions to:
receive a copy of network data received over at least one of the plurality of port interfaces and directed to a host device;
apply the one or more filters to the copy of the network data to obtain filtered network data, wherein the filtered network data comprises network packets directed to the host device, wherein at least some of the network packets are directed to the host device from another device;
sample at least a portion of the network packets;
generate one or more flow records comprising information about the sampled portion of the network packets, the information comprising at least a destination port, and the one or more flow records representing behavior of network traffic directed to the host device;
extract a plurality of features from the one or more flow records;
determine, using a machine learning (ML) detection system implemented by the acceleration hardware engine, that the host device is subject to a distributed denial of service (DDoS) attack based on the plurality of features; and
provide an alert of the DDoS attack to the host device in response to a determination that the host device is subject to the DDoS attack, the alert to cause an action associated with an enforcement rule to be performed at the host device.
2 . The switch of claim 1 , wherein the ML detection system comprises a tree-based ML model trained to determine whether the host device is subject to the DDoS attack based on the plurality of features.
3 . The switch of claim 1 , wherein the ML detection system comprises a regression model trained to predict whether the host device is subject to the DDoS attack based on the plurality of features.
4 . The switch of claim 1 , wherein the CPU is to send the enforcement rule to the host device in response to a determination that the host device is subject to the DDoS attack.
5 . The switch of claim 1 , wherein the CPU is to provide the alert to a data processing unit (DPU) of the host device, and wherein the alert is to cause the DPU to perform the action associated with the enforcement rule.
6 . The switch of claim 1 , wherein the CPU is to provide the alert to a hypervisor executed by the host device, the host device comprising a network interface card (NIC), and wherein the alert is to cause the hypervisor to perform the action associated with the enforcement rule.
7 . The switch of claim 1 , wherein the plurality of features comprises at least one of a source media access control (MAC) address, a destination MAC address, a source internet protocol (IP) address, a destination IP address, a source port, the destination port, a protocol identifier, a packet size, or a maximum packet size.
8 . The switch of claim 1 , further comprising:
network monitoring logic to extract the plurality of features and send the plurality of features to the ML detection system to determine whether the host device is subject to the DDoS attack.
9 . The switch of claim 8 , wherein the acceleration hardware engine is to:
receive the plurality of features from the network monitoring logic;
determine whether the host device is subject to the DDoS attack using a regression model trained to predict whether the host device is subject to the DDoS attack based on the plurality of features; and
send the alert to the host device responsive to the determination that the host device is subject to the DDoS attack.
10 . The switch of claim 8 , wherein the acceleration hardware engine is to:
receive the plurality of features from the network monitoring logic;
determine whether the host device is subject to the DDoS attack using a tree-based ML model trained to determine whether the host device is subject to the DDoS attack based on the plurality of features; and
send the alert to the host device responsive to the determination that the host device is subject to the DDoS attack.
11 . The switch of claim 1 , wherein the information further comprises at least one of a source port, a source address, a destination address, a protocol type, or a packet counter.
12 . A method comprising:
storing, at a switch, a configuration file that specifies one or more filters that extract for inclusion or remove from inclusion specified types of data;
receiving, at a central processing unit (CPU) of the switch, a copy of network data received over at least one of a plurality of port interfaces of the switch and directed to a host device;
applying the one or more filters to the copy of the network data to obtain filtered network data, wherein the filtered network data comprises network packets directed to the host device, wherein at least some of the network packets are directed to the host device from another device;
sampling at least a portion of the network packets;
generating one or more flow records comprising information about the sampled portion of the network packets, the information comprising at least a destination port, and the one or more flow records representing behavior of network traffic directed to the host device;
extracting a plurality of features from the one or more flow records;
determining, using a machine learning (ML) detection system implemented by an acceleration hardware engine of the switch, whether the host device is subject to a distributed denial of service (DDoS) attack based on the plurality of features; and
causing an action associated with an enforcement rule to be performed at the host device on subsequent network traffic data directed to the host device from the other device, responsive to a determination that the host device is subject to the DDoS attack.
13 . The method of claim 12 , wherein the information further comprises at least one of a source port, a source address, a destination address, a protocol type, or a packet counter.
14 . The method of claim 12 , wherein causing the action to be performed comprises sending an alert to a hypervisor executed by the host device, and wherein the hypervisor is to perform the action associated with the enforcement rule.
15 . The method of claim 12 , wherein the plurality of features comprises at least one of a source media access control (MAC) address, a destination MAC address, a source internet protocol (IP) address, a destination IP address, a source port, the destination port, a protocol identifier, a packet size, or a maximum packet size.
16 . The method of claim 12 , wherein causing the action to be performed comprises sending an alert to a data processing unit (DPU) of the host device, and wherein the DPU is to perform the action associated with the enforcement rule.
17 . The method of claim 16 , wherein determining whether the host device is subject to the DDoS attack comprises using a tree-based ML model trained to determine whether the host device is subject to the DDoS attack based on the plurality of features.
18 . The method of claim 16 , wherein determining whether the host device is subject to the DDoS attack comprises using a regression model trained to predict whether the host device is subject to the DDoS attack based on the plurality of features.
19 . A network device comprising:
a network interface;
a host interface;
an acceleration hardware engine;
a processing device operatively coupled to the network interface and the host interface, wherein the processing device is to:
receive a copy of network data received over the network interface and directed to a host device coupled to the host interface;
apply one or more filters to the copy of the network data to obtain filtered network data, wherein the filtered network data comprises network packets directed to the host device, wherein at least some of the network packets are directed to the host device from another device;
sample at least a portion of the network packets;
generate one or more flow records comprising information about the sampled portion of the network packets, the information comprising at least a destination port, and the one or more flow records representing behavior of network traffic directed to the host device;
extract a plurality of features from the one or more flow records;
determine, using a machine learning (ML) detection system implemented by the acceleration hardware engine, that the host device is subject to a distributed denial of service (DDoS) attack based on the plurality of features; and
provide an alert of the DDoS attack to the host device in response to a determination that the host device is subject to the DDoS attack, the alert to cause an action associated with an enforcement rule to be performed at the host device.
20 . The network device of claim 19 , wherein the network device is a network interface card (NIC) coupled to the host device.
21 . The network device of claim 19 , wherein the network device is a switch coupled to the host device.
22 . The network device of claim 19 , wherein the information further comprises at least one of a source port, a source address, a destination address, a protocol type, or a packet counter.