Reassembly-free deep packet inspection on multi-core hardware

Abstract

Some embodiments of reassembly-free deep packet inspection (DPI) on multi-core hardware have been presented. In one embodiment, a set of packets of one or more files is received at a networked device from one or more connections. Each packet is scanned using one of a set of processing cores in the networked device without buffering the one or more files in the networked device. Furthermore, the set of processing cores may scan the packets substantially concurrently.

Description

TECHNICAL FIELD

The present invention relates to intrusion detection and prevention in a networked system, and more particularly, to performing multiple packet payloads analysis on multi-core hardware.

BACKGROUND

Today, in many security products, scanning by pattern matching is used to prevent many types of security attacks. For example, some existing desktop virus scanning may include scanning files against certain recognizable patterns. These files may come from mail attachments or website downloads. These desktop applications are simpler in that by the time the pattern matching is performed, the input has been all accumulated in the correct order. The situation is more complicated for gateway products, such as firewalls, attempting to match patterns for other purposes. Some of these products scan for patterns over Transport Control Protocol (TCP) packets. Since TCP usually breaks down application data into chunks called TCP segments, the full pattern may reside in several TCP segments. One conventional approach is to reassemble all TCP packets together into one large chunk and perform pattern matching on this chunk, similar to scanning files. The disadvantage of this approach is that this approach requires processing to reassemble, and it further requires memory to buffer the intermediate result before pattern matching can take place.

To further complicate the problem, many security attacks exhibit more than one pattern, and thus, multiple pattern matching has to be performed in order to successfully screen out these attacks. Such a collection of patterns is called a signature. For example, an attack signature may contain a recognizable header and a particular phrase in the body. To detect such an attack, the detection mechanism has to match all the patterns in the signature. If only part of the signature is matched, false positives may occur. As such, the term “attack pattern” is used to refer to a single pattern or a signature.

When such attacks are transported over TCP, the contents, and therefore the recognizable patterns, may exist in different TCP segments. In fact, even a single pattern is often split over several segments. Therefore, two problems have to be solved at the same time. On one hand, the detection mechanism has to scan each pattern across multiple segments, and on the other hand, the detection mechanism also has to scan across patterns. One existing approach is to reassemble all packets and scan for each pattern in sequence. This approach is inefficient in terms of processing time and memory usage because scanning cannot start until all packets are received and reassembled and extra memory is needed to store the packets received.

Another problem in pattern matching is that the packets may arrive out of order. Again, using TCP as an example, the application data is broken into what TCP considers the best sized chunks to send, called a TCP segment or a TCP packet. When TCP sends a segment, it maintains a timer and waits for the other end to acknowledge the receipt of the segment. The acknowledgement is commonly called an ACK. If an ACK is not received for a particular segment within a predetermined period of time, the segment is retransmitted. Since the IP layer transmits the TCP segments as IP datagrams and the IP datagrams can arrive out of order, the TCP segments can arrive out of order as well. Currently, one receiver of the TCP segments reassembles the data so that the application layer receives data in the correct order.

An existing Intrusion Detection/Prevention System (IPS) typically resides between the two ends of TCP communication, inspecting the packets as the packets arrive at the IPS. The IPS looks for predetermined patterns in the payloads of the packets. These patterns are typically application layer patterns. For example, the pattern might be to look for the word “windows.” However, the word may be broken into two TCP segments, e.g., “win” in one segment and “dows” in another segment. If these two segments arrive in the correct order, then IPS can detect the word. However, if the segments arrive out of order, which happens relatively often, then the IPS may first receive the segment containing “dows”, and have to hold this segment and wait for the other segment. A typical approach is for the IPS to force the sender to re-transmit all the segments from the last missing one, hoping that the segments may arrive in order the second time. One disadvantage of this approach is the additional traffic in between and the additional processing on both ends of the TCP communication.

To take advantage of the introduction of multi-core processors (e.g., Intel® Core™2 Quad Processors from Intel Corporation of Santa Clara, Calif.), some conventional ISPs use multi-core processors to scan incoming segments to speed up the process. In general, each multi-core processor has two or more processing cores. According to one conventional approach, one of the processing cores is used to completely reassemble the file while the remaining processing cores perform scanning or pattern matching in the background after the file has been completely reassembled. However, this approach does not scale in terms of having enough memory to store all files. Also, background scanning by multiple processing cores is less efficient due to extra memory copying overhead and extra scheduling processing overhead.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which:

FIG. 1A illustrates one embodiment of a method to perform multiple packet analysis on multi-core hardware.

FIG. 1B illustrates an alternate embodiment of a method to perform multiple packet analysis on multi-core hardware.

FIG. 1C illustrates one embodiment of a method to perform deep packet inspection.

FIG. 2 illustrates an exemplary Deterministic Finite Automaton (DFA) according to one embodiment of the invention.

FIG. 3 illustrates a functional block diagram of one embodiment of multi-core hardware usable to perform multiple packet analysis.

FIG. 4 illustrates one embodiment of a system in which embodiments of the present invention may be implemented.

FIG. 5 illustrates a block diagram of an exemplary computer system, in accordance with one embodiment of the present invention.

DETAILED DESCRIPTION

Described herein are some embodiments of reassembly-free deep packet inspection on multi-core hardware. In one embodiment, a set of packets of one or more files is received at a networked device from one or more connections. Each packet is scanned using one of a set of processing cores in the networked device without buffering the one or more files in the networked device. Furthermore, the set of processing cores may scan the packets substantially concurrently.

In the following description, numerous details are set forth. It will be apparent, however, to one skilled in the art, that the present invention may be practiced without these specific details. In some instances, well-known structures and devices are shown in block diagram form, rather than in detail, in order to avoid obscuring the present invention.

Some portions of the detailed descriptions below are presented in terms of algorithms and symbolic representations of operations on data bits within a computer memory. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, and generally, conceived to be a self-consistent sequence of steps leading to a desired result. The steps are those requiring physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. It has proven convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like.

It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as “processing” or “computing” or “calculating” or “determining” or “displaying” or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.

The present invention also relates to apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer-readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, and magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, flash memory, magnetic or optical cards, or any type of media suitable for storing electronic instructions, and each coupled to a computer system bus.

The algorithms and displays presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform the required method steps. The required structure for a variety of these systems will appear from the description below. In addition, the present invention is not described with reference to any particular programming language. It will be appreciated that a variety of programming languages may be used to implement the teachings of the invention as described herein.

FIG. 1A illustrates one embodiment of a method to perform multiple packet analysis on multi-core hardware, where multiple processing cores of a set of processing cores are allowed to handle packets from the same connection (hereinafter, “connection X”). In some embodiments, the set of processing cores includes processing cores of a multi-core processor. The method may be performed by processing logic that may comprise hardware (e.g., circuitry, dedicated logic, programmable logic, processing cores, etc.), software (such as instructions run on a processing core), firmware, or a combination thereof.

Initially, one of a set of processing cores receives a packet from connection X (processing block 110). The packet is part of a file, which may be re-constructed by re-assembling the packet with other packets of the file. Then the processing core determines if the packet is in-order (processing block 112). For example, the processing core may check a sequence number in a header of the packet against a next packet sequence number of connection X, which may be stored in a database commonly accessible by the processing cores.

If the packet is not in-order, i.e., out-of-order, then the processing core may buffer the packet in an out-of-order buffer associated with connection X (processing block 114). The processing core may allow the packet to pass (processing block 115). Then the processing core waits for another new incoming packet (processing block 120).

If the packet is in-order, then the processing core performs deep packet inspection (DPI) on the packet (processing block 116). Details of some embodiments of DPI are discussed below. Then the processing core checks if there is any packet in the out-of-order buffer associated with connection X that recently became in-order (processing block 118). If there is no packet in the out-of-order buffer associated with connection X that is next in sequence (in-order), the processing core transitions to processing block 120 to wait for another new incoming packet. Otherwise, if there is a packet in the out-of-order buffer associated with connection X that is now in-order, then the processing core removes this packet and performs DPI on this packet (processing block 122). When the processing core completes DPI on this packet, the processing core returns to processing block 118 to check if there is another packet in the out-of-order buffer associated with connection X that is in-order.

Note that the incoming packets are scanned without buffering the file for reassembly because the packets can be inspected for the predetermined pattern without being reassembled into the file. Thus, the above technique is well suited for IPSs that have limited capacity for buffering or storage. Furthermore, the above technique allows the set of processing cores to scan incoming packets substantially concurrently. Therefore, the speed of the scanning may be improved over conventional approaches.

FIG. 1B illustrates one embodiment of a method to perform multiple payload analysis on multi-core hardware, where only a single core in a set of processing cores is allowed to handle packets from a particular connection (hereinafter, “connection X”) at a time. In some embodiments, the set of processing cores includes processing cores of a multi-core processor. The method may be performed by processing logic that may comprise hardware (e.g., circuitry, dedicated logic, programmable logic, processing cores, etc.), software (such as instructions run on a processing core), firmware, or a combination thereof.

Initially, one processing core of the set of processing cores receives a packet from connection X (processing block 130). Then the processing core checks if there is another processing core in the set of processing cores handling another packet from connection X (processing block 132). If there is another processing core handling another packet from connection X currently, then the processing core postpones handling of this packet until the other processing core is done with the other packet from connection X (processing block 134). The processing core may transition to processing block 144 to wait for another new incoming packet.

If the processing core determines that there is no other processing core in the set of processing cores handling another packet from connection X, then the processing core checks if this packet is in-order (processing block 136). If this packet is not in-order, i.e., out-of-order, then the processing core buffers this packet in an out-of-order buffer associated with connection X (processing block 140). The processing core may allow this packet to pass (processing block 142). Then the processing core waits for another new incoming packet (processing block 144).

If the processing core determines that this packet is in-order, then the processing core performs DPI on this packet (processing block 138). Details of some embodiments of DPI are discussed below. After performing DPI on the packet, the processing core checks if there is any packet in the out-of-order buffer associated with connection X, which is now in-order (processing block 146). If there is a packet in the out-of-order buffer that is now in-order, then the processing core removes the packet that recently became in-order from the out-of-order buffer and performs DPI on this packet (processing block 148). Then the processing core returns to processing block 146 to repeat the above process. If there is no packet in the out-of-order buffer that is in-order, then the processing core transitions to processing block 144 to wait for another new incoming packet.

Like the technique illustrated in FIG. 1A, the technique illustrated in FIG. 1B also allows scanning of the incoming packets without buffering the file for reassembly because the packets can be scanned for the predetermined pattern, without reassembling the packets into the file, by DPI.

FIG. 1C illustrates one embodiment of a method to perform deep packet inspection (DPI) using one of a set of processing cores. In some embodiments, the set of processing cores includes processing cores of a multi-core processor. The method may be performed by processing logic that may comprise hardware (e.g., circuitry, dedicated logic, programmable logic, processing cores, etc.), software (such as instructions run on a processing core), firmware, or a combination thereof.

Initially, the processing core starts DPI on a packet from connection X at block 150. This packet is hereinafter referred to as the current packet. The processing core performs pattern matching on the current packet from the last stored state of pattern matching for connection X (processing block 152). Specifically, the processing core is trying to look for a predetermined pattern or signature in the incoming packets, which may be associated with a computer virus or malicious code. By identifying such pattern or signature in the incoming packets and blocking at least one of the packets containing part of the predetermined pattern or signature, the set of processing cores can protect a system from computer viral attack. In some embodiments, the last stored state of pattern matching for connection X is stored in a database commonly accessible by the set of processing cores. As such, each of the set of processing cores can handle packets from connection X, even though some of the packets may be inspected by different processing cores.

In some embodiments, if there is a match between a predetermined pattern and the data pattern in the incoming packets inspected so far (which includes the current packet), then the processing core blocks the current packet (processing block 160). Then the processing core may issue an alarm to warn a system administrator of detection of potentially malicious code or virus in the incoming packets (processing block 162), and the process ends at block 164.

If there is no match between the predetermined pattern and the data pattern in the incoming packets inspected so far, then the processing core may update and store the current state of pattern matching of connection X in the database (processing block 156). The method then ends at block 158.

In some embodiments, pattern matching performed in DPI is accomplished using Deterministic Finite Automaton (DFA). An exemplary DFA is shown in FIG. 2 to illustrate the concept.

FIG. 2 illustrates an exemplary DFA according to one embodiment of the invention. In this example, an IPS is programmed to detect and to prevent a pattern of “0111” to pass through. The DFA 200 shown in FIG. 2 corresponds to this pattern. A set of processing cores may use the DFA 200 to perform pattern matching on a number of packets to determine whether the packets contain the pattern “0111”. Furthermore, to simplify the illustration, it is assumed in this example that each packet contains only one digit. However, it should be appreciated that the concept is applicable to scenarios where a packet contains more than one digits and/or alphabetic letters.

Referring to FIG. 2, the DFA 200 includes 5 states 211-219. The states 211-219 in the DFA 200 may be referred to as nodes. A processing core in the set of processing cores begins pattern matching at the initial state 211. If a packet received contains a “1”, the processing core remains in the initial state 211. If the packet contains a “0”, which corresponds to the first digit in the predetermined pattern, the processing core transitions to the A state 213.

If the processing core receives a “0” subsequently, the processing core remains in the A state 213. If the processing core receives a “1”, which corresponds to the second digit in the predetermined pattern, then the processing core transitions into the B state 215. From the B state 215, the processing core may transition back to the A state 213 if the next packet received contains a “0”. If the next packet received contains a “1”, which corresponds to the third digit in the predetermined pattern, then the processing core transitions to the C state 217. However, note that another processing core in the set of processing cores may receive and process the next packet in some embodiments.

From the C state 217, the processing core may transition back to the A state 213 if the next packet received contains a “0”. If the next packet received contains a “1”, which corresponds to the last digit in the predetermined pattern, then the processing core transitions to the final state 219. When the processing core reaches the final state 219, the processing core knows that the packets received so far contain the predetermined pattern. Hence, the processing core may perform the appropriate operations in response to receiving the predetermined pattern, such as blocking the packet of the predetermined pattern last received and issuing an alarm to alert system administrators. To keep track of which state of the DFA is in currently, the processing core stores the current state of the DFA in a database commonly accessible by the set of processing cores. As such, another processing core may continue pattern matching on the next packet from the current state if the other processing core receives the next packet. Furthermore, the current state of the DFA may be associated with a connection from which the packet is received so that the set of processing cores may inspect packets from multiple connections using the information from the database.

One advantage of using the DFA to perform pattern matching on packets is to eliminate the need to reassemble the packets because the processing cores can walk through the DFA as each packet is received and examined. Because a pattern is typically broken up into a number of segments and each segment is transmitted using a packet, it is necessary to inspect multiple packets in order to identify the pattern. Using the DFA, the processing cores may not have to reassemble the packets in order to find out what the pattern contained in the packets is in order to match the pattern against a predetermined pattern. The processing cores may perform pattern matching on a packet-by-packet basis as each of the packets is received without reassembling the packets by walking through the DFA. If a processing core reaches a final state, there is a match between the pattern contained in the packets received so far and the predetermined pattern. There is no need to store the packets for reassembling the packets. Instead, the processing cores may simply store the current state of the DFA in a database commonly accessible by the processing cores.

The concept described above may be expanded to signature detection. A signature is a collection of multiple patterns. To keep track of which pattern within a signature is being matched, processing logic may use a tree structure, where each node within the tree structure corresponds to a pattern and each pattern is represented using a DFA. Alternatively, a single DFA may represent multiple patterns.

FIG. 3 illustrates a functional block diagram of one embodiment of multi-core hardware usable to perform multiple payload analysis in an IPS. The IPS may be implemented within a set-top box coupled to a protected network. The multi-core hardware 300 includes a set of processing cores 310, a pattern matching database 320, and an out-of-order buffer 330. In some embodiments, the set of processing cores 310 includes processing cores in a multi-core processor. The processing cores 310 are communicably coupled to the database 320 so that each of the processing cores 310 may retrieve and update information in the database 320. Likewise, the processing cores 310 are also communicably coupled to the out-of-order buffer 330 so that each of the processing cores 310 may access the out-of-order buffer 330.

In some embodiments, the processing cores 310 receive packets from one or more connections. To prevent harmful virus or malicious code from reaching the protected network, the processing cores 310 performs reassembly-free DPI on the packets. When one of the processing cores 310 receives a packet, the processing core may determine if the packet is in-order or out-of-order. An out-of-order packet may be temporarily stored in the out-of-order buffer 330 and be associated with the connection from which the out-of-order packet is received. In-order packets are examined by the processing cores 310 and are allowed to pass to the protected network if no pattern of harmful virus or malicious code is detected. The processing cores 310 update and store the current pattern matching state of each connection in the database 320. As such, any one of the processing cores 310 can continue with the on-going pattern matching from the current state of a connection that sends the current packet. In some embodiments, the database 320 includes a relational database that stores the current pattern matching states 324 with their corresponding connections 322 as shown in FIG. 2. Details of some embodiments of the method to perform reassembly-free DPI have been discussed above.

FIG. 4 illustrates one embodiment of a system in which embodiments of the present invention may be implemented. The system 400 includes a client machine 412 within a protected network 410, an IPS 420, and a network 430. The protected network 410 is communicably coupled to the network 430 via the IPS 420. Thus, packets transmitting between the protected network 410 and the network 430 have to pass through the IPS 420. In some embodiments, there may be more than one client machines coupled to the protected network 410. The network 430 may include a variety of networks, such as local area network (LAN), wide area network (WAN), etc. Furthermore, the network 430 may be publicly accessible, and therefore, computer virus and malicious code targeting the protected network 410 may be sent from the network 430. As such, the IPS 420 scans the incoming packets to prevent computer virus and malicious code from entering the protected network 410.

In some embodiments, the IPS 420 includes a multi-core processor 421, an out-of-order buffer 423, and a pattern matching database 425. The multi-core processor 421 includes a set of processing cores, such as the processing cores 310 shown in FIG. 3.

In some embodiments, each of the processing cores receives packets from the network 430 through different connections. Furthermore, the packets may arrive out-of-order, and if so, the out-of-order packets may be temporarily stored in the out-of-order buffer 423 to be inspected later. The processing cores of the multi-core processor 421 perform DPI on the in-order packets and store the current pattern matching states of the connections in the pattern matching database 425. If a pattern associated with computer virus or malicious code is identified in the incoming packets inspected so far, the multi-core processor 421 blocks the packet currently being inspected and may further issue a warning to a system administrator. If no pattern associated with computer virus or malicious code is identified in the incoming packets inspected so far, then the multi-core processor 421 allows the packet currently being inspected to pass to the protected network 410, which may be further transmitted to the client machine 412. By blocking the packet currently being inspected if the pattern is identified in the packets received so far, the computer virus or malicious code cannot be completely passed into the protected network 410, and hence, the computer virus or malicious code cannot be completely reassembled on the client machine 412. The incomplete computer virus or malicious code typically cannot harm the client machine 412 coupled thereto. Details of some embodiments of a method to perform reassembly-free DPI have been discussed above.

FIG. 5 illustrates a diagrammatic representation of a machine in the exemplary form of a computer system 500 within which a set of instructions, for causing the machine to perform any one or more of the methodologies discussed herein, may be executed. In alternative embodiments, the machine may be connected (e.g., networked) to other machines in a LAN, an intranet, an extranet, and/or the Internet. The machine may operate in the capacity of a server or a client machine in client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine may be a personal computer (PC), a tablet PC, a set-top box (STB), a Personal Digital Assistant (PDA), a cellular telephone, a web appliance, a server, a network router, a switch or bridge, or any machine capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that machine. Further, while only a single machine is illustrated, the term “machine” shall also be taken to include any collection of machines that individually or jointly execute a set (or multiple sets) of instructions to perform any one or more of the methodologies discussed herein.

The exemplary computer system 500 includes a processing device 502, a main memory 504 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM (RDRAM), etc.), a static memory 506 (e.g., flash memory, static random access memory (SRAM), etc.), and a data storage device 518, which communicate with each other via a bus 532.

Processing device 502 represents one or more general-purpose processing devices such as a microprocessor, a central processing unit, or the like. More particularly, the processing device may be complex instruction set computing (CISC) microprocessor, reduced instruction set computing (RISC) microprocessor, very long instruction word (VLIW) microprocessor, or processor implementing other instruction sets, or processors implementing a combination of instruction sets. Processing device 502 may also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), network processor, or the like. The processing device 502 is configured to execute the processing logic 526 for performing the operations and steps discussed herein.

The computer system 500 may further include a network interface device 508. The computer system 500 also may include a video display unit 510 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 512 (e.g., a keyboard), a cursor control device 514 (e.g., a mouse), and a signal generation device 516 (e.g., a speaker).

The data storage device 518 may include a machine-accessible storage medium 530 (also known as a machine-readable storage medium or a computer-readable medium) on which is stored one or more sets of instructions (e.g., software 522) embodying any one or more of the methodologies or functions described herein. The software 522 may also reside, completely or at least partially, within the main memory 404 and/or within the processing device 502 during execution thereof by the computer system 500, the main memory 504 and the processing device 502 also constituting machine-accessible storage media. The software 522 may further be transmitted or received over a network 520 via the network interface device 508.

While the machine-accessible storage medium 530 is shown in an exemplary embodiment to be a single medium, the term “machine-accessible storage medium” should be taken to include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) that store the one or more sets of instructions. The term “machine-accessible storage medium” shall also be taken to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by the machine and that cause the machine to perform any one or more of the methodologies of the present invention. The term “machine-accessible storage medium” shall accordingly be taken to include, but not be limited to, solid-state memories, optical and magnetic media, etc. In some embodiments, machine-accessible storage medium may also be referred to as computer-readable storage medium.

Thus, some embodiments of reassembly-free DPI on multi-core hardware have been described. It is to be understood that the above description is intended to be illustrative, and not restrictive. Many other embodiments will be apparent to those of skill in the art upon reading and understanding the above description. The scope of the invention should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. A method comprising: receiving a plurality of packets of one or more files at a networked device comprising a plurality of processing cores, the packets from a plurality of connections;processing each of the received packets, wherein processing each received packet comprises: determining from which of the plurality of connections the packet came, andpostponing processing one of the received packets based on a determination that another processing core is currently processing a packet from a same connection, andcontinuing to process the one of the received packets based on a determination that no other processing core is currently processing a packet from the same connection;storing a current state of pattern matching in a database in memory accessible to each of the plurality of processing cores, wherein the current state of pattern matching corresponds to packets received from the determined corresponding connection, and wherein a plurality of other current states of pattern matching are stored for other connections from the plurality of connections;scanning each of the plurality of packets using one of the plurality of processing cores in the networked device without buffering the one or more files in the networked device, such that the plurality of processing cores scan the plurality of packets substantially concurrently, wherein when the plurality of packets are from one of the plurality of connections, a first processing core of the plurality of processing core receives an in-order packet and scans the in-order packet, a second processing core of the plurality of processing core receives an out-of-order packet and temporarily buffers the out-of-order packet in an out-of-order buffer without scanning the out-of-order packet, wherein the first processing core retrieves a next in order packet from the out-of-order buffer to scan after scanning the in-order packet; andupdating the current state of pattern matching based on a plurality of scan results from the plurality of processing cores, the updated current state of pattern matching stored with the determined corresponding connection.

2. The method of claim 1, further comprising: resolving conflicts between out-of-order packets among the plurality of packets.

3. The method of claim 2, wherein when the plurality of packets are from distinct ones of the plurality of connections, the first processing core of the plurality of processing cores receives a first packet from a first connection, and resolving conflicts between out-of-order packets further comprises: the first processing core determining if the first packet is in-order or out-of order;the first processing core scanning the packet if the first packet is in-order; andthe first processing core temporarily buffering the first packet in an out-of order buffer associated with the first connection without scanning the first packet if the first packet is out-of-order.

4. The method of claim 3, wherein resolving conflicts between out-of-order packets further comprises: the second processing core of the plurality of processing cores receiving a second packet from the first connection while the first processing core is still processing the first packet; andthe second processing core re-scheduling scanning of the second packet to a later time.

5. The method of claim 1, wherein storing the current state of pattern matching further comprises storing a plurality of current states in the database, each current state associated with one of the plurality of connections.

6. An apparatus comprising: a network interface to receive a plurality of packets of one or more files from a plurality of connections;a plurality of processing cores to perform reassembly-free deep packet inspection on the plurality of packets without buffering the one or more files such that the plurality of processing cores scan the plurality of packets substantially concurrently, wherein each processing core processes each received packet by: determining from which of the plurality of connections the packet came,postponing processing one of the received packets based on a determination that another processing core is currently processing a packet from a same connection, andcontinuing to process the one of the received packets based on a determination that no other processing core is currently processing a packet from the same connection, wherein when the plurality of packets are from one of the plurality of connections, a first processing core of the plurality of processing core receives an in-order packet and scans the in-order packet, a second processing core of the plurality of processing core receives an out-of-order packet and temporarily buffers the out-of-order packet in an out-of-order buffer without scanning the out-of-order packet, wherein the first processing core retrieves a next in order packet from the out-of-order buffer to scan after scanning the in-order packet; andmemory accessible to each of the plurality of processing cores, the memory associated with a database for storing a current state of pattern matching, the current state of pattern matching corresponding to packets from the determined corresponding connection, wherein a plurality of other current states of pattern matching are stored for other connections from the plurality of connections, wherein the current state of pattern matching is updated based on a plurality of scan results from the plurality of processing cores, the updated current state of pattern matching stored with the determined corresponding connection.

7. The apparatus of claim 6, wherein the plurality of processing cores resolve conflicts between out-of-order packets among the plurality of packets.

8. The apparatus of claim 6, wherein when the plurality of packets are from distinct ones of the plurality of connections, the first processing core of the plurality of processing cores receives a first packet from a first connection and scans the packet if the first packet is in-order, and temporarily buffers the first packet in an out-of-order buffer associated with the first connection without scanning the first packet if the first packet is out-of-order.

9. The apparatus of claim 6, wherein the second processing core to receive a second packet from a first connection while the first processing core is still processing the first packet, and to re-schedule scanning of the second packet to a later time.

10. The apparatus of claim 6, wherein the database further stores a plurality of current states, each current state associated with one of the plurality of connections.

11. A system comprising the apparatus of claim 6, further comprising: one or more client devices coupled to receive the plurality of packets after the plurality of packets have been scanned without identifying any prohibited content.

12. A non-transitory computer-readable medium embodying instructions that, when executed by a processor, will cause the processor to perform operations comprising: receiving a plurality of packets of one or more files at a networked device comprising a plurality of processing cores, the packets from plurality of connections;processing each of the received packets, wherein processing each received packet comprises: determining from which of the plurality of connections the packet came,postponing processing one of the received packets based on a determination that another processing core is currently processing a packet from a same connection, andcontinuing to process the one of the received packets based on a determination that no other processing core is currently processing a packet from the same connection;storing a current state of pattern matching in a database accessible to each of the plurality of processing cores, wherein the current state of pattern matching corresponds to packets from the determined corresponding connection, and wherein a plurality of other current states of pattern matching are stored for other connections from the plurality of connections;scanning each of the plurality of packets using one of the plurality of processing cores in the networked device without buffering the one or more files in the networked device, such that the plurality of processing cores scan the plurality of packets substantially concurrently, wherein when the plurality of packets are from one of the plurality of connections, a first processing core of the plurality of processing core receives an in-order packet and scans the in-order packet, a second processing core of the plurality of processing core receives an out-of-order packet and temporarily buffers the out-of-order packet in an out-of-order buffer without scanning the out-of-order packet, wherein the first processing core retrieves a next in order packet from the out-of-order buffer to scan after scanning the in-order packet; andupdating the current state of pattern matching based on a plurality of scan results from the plurality of processing cores, the updated current state of pattern matching stored with the determined corresponding connection.

13. The non-transitory computer-readable medium of claim 12, wherein the operations further comprise: resolving conflicts between out-of-order packets among the plurality of packets.

14. The non-transitory computer-readable medium of claim 12, wherein when the plurality of packets are from distinct ones of the plurality of connections, the first processing core of the plurality of processing cores receives a first packet from a first connection, and resolving conflicts between out-of-order packets further comprises: the first processing core determining if the first packet is in-order or out-of order;the first processing core scanning the packet if the first packet is in-order; andthe first processing core temporarily buffering the first packet in an out-of order buffer associated with the first connection without scanning the first packet if the first packet is out-of-order.

15. The non-transitory computer-readable medium of claim 14, wherein resolving conflicts between out-of-order packets further comprises: the second processing core of the plurality of processing cores receiving a second packet from the first connection while the first processing core is still processing the first packet; andthe second processing core re-scheduling scanning of the second packet to a later time.

16. The non-transitory computer-readable medium of claim 12, wherein storing the current state of pattern matching further comprises: storing a plurality of current states in the database, each current state associated with one of the plurality of connections.