Imported: 13 Feb '17 | Published: 18 Jan '11
USPTO - Utility Patents
The invention prevents server overload and possible server crippling due to a flooding of connectionless datagrams caused by intentional attack or otherwise. In response to a datagram from a host for a specified port, the number of datagrams already queued to the port from the host is determined. If this number exceeds a first threshold, the datagram is discarded. In the preferred embodiment, the threshold is determined by multiplying a percentage P by the number of available queue slots remaining for the port.
The invention relates generally to the field of networking and specifically to defending against attacks by malicious users attempting to disable a server by flooding the server with connectionless datagrams.
Flooding attacks have recently been used with Increasing frequency to target and disable servers on the Internet. A flooding attack occurs when a user sends a large number of requests to a server in a relatively short period of time with an intent to overload and thereby disable the server. A flood of packets from a malicious user can overload a server in the same way that a flood of packets from a misconfigured system can overload a server. But the end result is the same; the server becomes overloaded in trying to service the requests. This prevents legitimate requests from being timely served and often disables a server or causes it to crash. A number of flooding attacks have been reported in the news recently on some well known web sites. These attacks were characterized by a flood of individual connection requests to establish initial communications. A related patent application Ser. No. 09/502,478 filed Feb. 11, 2000, discloses an algorithm to defend against such connection request attacks. However, it is also possible to attack a server by flooding it with connectionless datagrams, such as might occur in the UDP (user datagram) protocol. The effect is essentially the same; the server becomes overloaded in trying to service the horde of datagrams and can even become totally disabled. Flooding attacks are very difficult for traditional intrusion detection systems to prevent due to the difficulty of determining whether the traffic is legitimate or not.
The invention recognizes that the consequences of intentional datagram flooding attacks and unintentional overload situations resulting from a burst of datagrams can be mitigated by dropping the traditional notion of attempting to distinguish between legitimate and illegitimate traffic. In the invention, all datagram traffic is subject to a policy that attempts to guarantee that legitimate work will be performed and a server will not crash in flooding situations, irrespective of whether the flooding is caused by legitimate or illegitimate datagram traffic. The invention helps to prevent a server from crashing due to overload and it prevents one or more attackers from consuming all server resources.
In response to the arrival of a datagram destined for a specified port on a server, the transmitting host is determined and the number of datagrams already queued for the same host and for the same port is determined. If this number exceeds a first threshold, the request for a connection is denied.
The first threshold is dynamically determined in the preferred embodiment. The owner of a server specifies for each port that is subject to datagram flooding checks a maximum number of queued datagrams (M) allowed at any time to the port and a controlling percentage (P) of available queue slots remaining for the port. The invention keeps track of the number (A) of queued datagrams for the port and it calculates the number of available queue slots (I) by subtracting the number of queued datagrams from the maximum number of datagrams (I=M−A). If the number of datagrams already queued for the transmitting host is equal to or greater than P times the number of queue slots left (=>P*I), then the present datagram is refused. Otherwise, the datagram is queued and the number of queued datagrams (A) for the port is incremented by one.
The maximum number of datagrams and the threshold percentage P will be difficult for most owners to configure. Therefore, a “statistics” mode is provided that measures normal traffic loads of different servers and suggests an appropriate maximum and threshold that will not hamper similar legitimate traffic loads. This statistics mode is not part of the claimed invention and is not described further herein.
The invention requires that an owner of a server that uses the invention configure the server with certain parameters for use by the invention. By way of example, the preferred embodiment requires that the owner specify for each port number subject to datagram flooding checks a maximum number of datagrams (M) allowed at any time to be queued to the port and a controlling percentage (P) of available queue slots remaining for the port. The percentage P is used to establish a threshold to trigger the denial to service a datagram. As datagrams are queued and serviced, the server dynamically maintains the number of available queue slots for each port that is subject to flooding checks.
An entry is made to step 100 in FIG. 1 when a datagram is received at a network server. The first step 106 determines from the datagram the port number to which the datagram is directed. The port number contained in a datagram represents a destination within a given host computer to which the datagram is to be delivered. Some ports are reserved for standard services. For example, the Network File System (NFS) is one example of a standard service that receives UDP datagrams.
The identity (the IP address) of the sending host is also determined from the datagram. The port number is used by step 108 to locate a memory control block for the port or to create one if a port control block does not presently exist. Attached to the port control block are a plurality of host control blocks for hosts that presently have one or more datagrams in queue. If the sending host does not have a host control block, one is created. A host control block contains, among other things, a count of the port connections presently assigned to the host.
Step 108 determines from the datagram the identity of the transmitting host that initiated the datagram and it uses the port number and the host identity to locate a memory control block or it creates a memory control block if one does not presently exist. An existing memory control block contains, among other things, a count of the number of datagrams presently queued by the host. Step 108 determines the port number to which this datagram is directed.
At step 110, the server fetches from the memory control block the maximum number of queued datagrams M specified for this port number, the controlling percentage P and the number A of queued datagrams. Step 112 calculates the number I of available queue slots as M−A. Step 114 determines if the number of datagrams already queued to the transmitting host is equal to or greater than P times I. If so, then the datagram is discarded and the queuing algorithm exits at 118. On the other hand, if the number of queued datagrams already initiated by the transmitting host is less than P times I, the datagram is queued for service at step 116 and A is incremented by one to update the number of datagrams in queue for this port number.
The computer program that has been described can be executed on virtually any type of computer, ranging from personal computers to large mainframes such as IBM's System 390 machines. The only requirement is that the computer is configured with network communication software and is accessible as a server via a network.
Skilled artisans in the fields to which the invention pertains will recognize that numerous variations can be made to the embodiments disclosed herein and still remain within the spirit and scope of the invention.