US 20060253903 A1
Methods and systems for firewall/data protection that filters data packets in real time and without packet buffering are disclosed. A data packet filtering hub, which may be implemented as part of a switch or router, receives a packet on one link, reshapes the electrical signal, and transmits it to one or more other links. During this process, a number of filters checks are performed in parallel, resulting in a decision about whether each packet should or should not be invalidated by the time that the last bit is transmitted. To execute this task, the filtering hub performs rules-based filtering on several levels simultaneously, preferably with a programmable logic or other hardware device. Various methods for packet filtering in real time and without buffering with programmable logic are disclosed. The system may include constituent elements of a stateful packet filtering hub, such as microprocessors, controllers, and integrated circuits. The system may be reset, enabled, disabled, configured, and/or reconfigured with toggles or other physical switches. Audio and visual feedback may be provided regarding the operation and status of the system.
1. A method for communicating data between an external computing system and an internal computing system over a packet-based network, comprising the steps of:
receiving a communication packet from the external computing system over the network, the packet having at least a first portion and an end portion, and transmitting the packet to the internal computing system;
in parallel with the step of receiving and transmitting the packet, determining characteristics of the packet from the first portion;
in parallel with the step of receiving and transmitting the packet, performing a plurality of checks on the packet, wherein at least certain of the plurality of checks are performing in parallel with other of the plurality of checks;
in parallel with the step of receiving and transmitting the packet, determining if the packet should be a valid packet or an invalid packet based on the plurality of checks; and
after receiving the end portion of the packet, selectively altering the end portion of the packet based on whether the packet has been determined to be a valid packet or an invalid packet, wherein the packet is selectively altered to be invalid if it was determined that the packet should be an invalid packet.
The present invention relates to computer security and data protection systems and methods, and more particularly to firewall and data protection systems and methods for filtering packets, such as from the Internet, in real time and without packet buffering.
The use of the Internet has exploded in recent years. Small and large companies as well as individual users are spending more time with their computers connected to the Internet. With the advent of Internet technologies, such as cable modems, digital subscriber lines, and other “broadband” access devices, users are connecting their computers to the Internet for extended periods of time.
Such extended or “persistent” connection to the Internet brings many advantages to users in immediate access to the content on the Internet through the use of email, search engines, and the like. Unfortunately, however, persistent access to the Internet exposes connected computers to potential security threats, where intruders and “hackers” may compromise proprietary systems, engage in information theft, or take control of the connected computers remotely. With more sophisticated tools at their disposal, hackers pose security and privacy risks to systems with persistent access to the Internet. Such security risks are even present for computers connected to the Internet for limited periods of time (such as through dial-up, modem connections), though to a lesser degree than the extended access computers.
There are currently many different types of firewall systems available on the market, including proxy servers, application gateways, stateful inspection firewalls, and packet filtering firewalls, each of which provides a variety of strategies and services for data protection. Conventional packet filters typically are computers, routers, or ASICs based on general purpose CPUs. They perform their filtering duties by receiving a packet, buffering the data until a determination can be made, and forwarding the packet as applicable for the particular system.
For example, a dual-homed, Linux-based filter with two network cards might receive a packet completely, evaluate whether it meets specific criteria, and transmit the packet on the other network card. In another example, a router designed for switch mode routing might begin buffering a packet until a decision is made, then forward the packet on the applicable interface while still receiving the packet. With most packet filters, software is used and data is buffered.
Sophisticated computer users working for medium- to large-sized companies have a variety of relatively expensive protection devices and tools at their disposal. Such devices and tools typically screen data packets received from the Internet with sophisticated software-based filtering techniques. Using relatively complex tools for software analysis, each packet is stored in a buffer and examined sequentially with software-based rules, which results in each packet being either accepted (and passed to the computer) or rejected (and disposed of by the software). This software often requires substantial computer knowledge and experience. Users of such devices and tools typically have an expertise in network administration or a similar field, so they can configure, optimize, and even build the complex filtering and security options provided by the software.
While such devices and tools can be quite effective in providing “firewall” protection for sophisticated users of large office systems, they pose several barriers to unsophisticated users of small office and home systems in the growing SOHO market. Current large office systems are expensive, difficult to set up, and require technical skills. What is needed for SOHO systems is a relatively inexpensive, uncomplicated, “plug and play” type of Internet protection system that can be easily connected and configured by relatively unsophisticated users.
In accordance with the present invention, devices, methods and systems are provided for the filtering of Internet data packets in real time and without packet buffering. A stateful packet filtering hub is provided in accordance with preferred embodiments of the present invention. The present invention also could be implemented as part of a switch or incorporated into a router.
A packet filter is a device that examines network packet headers and related information, and determines whether the packet is allowed into or out of a network. A stateful packet filter., however, extends this concept to include packet data and previous network activity in order to make more intelligent decisions about whether a packet should be allowed into or out of the network. An Ethernet hub is a network device that links multiple network segments together at the medium level (the medium level is just above the physical level, which connects to the network cable), but typically provides no capability for packet-type filtering. As is known, when a hub receives an Ethernet packet on one connection, it forwards the packet to all other links with minimal delay and is accordingly not suitable as a point for making filtering-type decisions. This minimum delay is important since Ethernet networks only work correctly if packets travel between hosts (computers) in a certain amount of time.
In accordance with the present invention as the data of a packet comes in from one link (port) the packet's electrical signal is reshaped and then transmitted down other links. During this process, however, a filtering decision is made between the time the first bit is received on the incoming port and the time the last bit is transmitted on the outgoing links. During this short interval a substantial number of filtering rules or checks are performed, resulting in a determination as to whether the packet should or should not be invalidated by the time that the last bit is transmitted. To execute this task, the present invention performs multiple filtering decisions simultaneously: data is received; data is transmitted; and filtering rules are examined in parallel and in real time. For example, on a 100 Mbit/sec Ethernet network, 4 bits are transmitted every 40 nano seconds (at a clock speed of 25 MHz). The present invention makes a filtering decision by performing the rules evaluations simultaneously at the hardware level, preferably with a programmable logic device.
The present invention may employ a variety of networking devices in order to be practical, reliable and efficient. In addition, preferred embodiments of the present invention may include constituent elements of a stateful packet filtering hub, such as microprocessors, controllers, and integrated circuits, in order to perform the real time, packet-filtering, without requiring buffering as with conventional techniques. The present invention preferably is reset, enabled, disabled, configured and/or reconfigured with relatively simple toggles or other physical switches, thereby removing the requirement for a user to be trained in sophisticated computer and network configuration. In accordance with preferred embodiments of the present invention, the system may be controlled and/or configured with simple switch activation(s).
Accordingly, one object of the present invention is to simplify the configuration requirements and filtering tasks of Internet firewall and data protection systems.
Another object is to provide a device, method and system for Internet firewall and data protection that does not require the use of CPU-based systems, operating systems, device drivers, or memory bus architecture to buffer packets and sequentially carry out the filtering tasks.
A further object of the present invention is to perform the filtering tasks of Internet firewall protection through the use of hardware components.
Another object is to utilize programmable logic for filtering tasks.
Still another object is to provide a device, method, and system to carry out bitstream filtering tasks in real time.
Yet another object is to perform parallel filtering, where packet data reception, filtering, and transmission are conducted simultaneously.
A further object of the present invention is to perform the filtering tasks relatively faster than current state-of-the-art, software-based firewall/data protection systems.
Another object is to provide a device, method and system for firewall protection without the use of a buffer or temporary storage area for packet data.
Still another object of the present invention is to design a device, method and system that does not require software networking configurations in order to be operational.
A further object of the present invention is to provide a device, method and system for Internet firewall and data security protection that supports partitioning a network between client and server systems.
It is a yet another object of the present invention to provide a device, method and system for Internet firewall and data protection that supports multiple networking ports.
Another object is to maintain stateful filtering support for standard data transmission protocols on a per port basis.
Still another object of is to configure network functionality using predefined toggles or other types of physical switches.
A further object of the present invention is to conduct packet filtering without requiring a MAC address or IP address to perform packet filtering.
Yet another object of the present invention is to facilitate the shortest time to carry out bitstream filtering tasks.
Finally, it is another object of the present invention to be able to perform filtering rules out of order and without the current state-of-the-art convention of prioritizing the filtering rules serially.
The present invention may be more fully understood by a description of certain preferred embodiments in conjunction with the attached drawings in which:
The present invention will be described in greater detail with reference to certain preferred and alternative embodiments. As described below, refinements and substitutions of the various embodiments are possible based on the principles and teachings herein.
With reference to
With reference to
Repeater core 16 functions as an Ethernet repeater (as defined by the network protocols of the IEEE standard 802.3) and serves to receive packets from external PHY 14, reshape the electrical signals thereof, and transmit the packets to internal PHY 18, which is coupled to internal network 20. While the packet is being received, reshaped, and transmitted between PHYs 14 and 18, however, it is simultaneously being evaluated in parallel with filtering rules to determine if it should be allowed to pass as a valid packet (as will be described in greater detail elsewhere herein). As with the discussion regarding the PHY interfaces and controllers, changes in networking standards may alter the components functionality (such as the characteristics of repeater core 16), but not the basic parallel, real time packet filtering in accordance with the present invention. (In an alternate embodiment, for example, the data protection system may use switch logic or router logic; in full duplex, the same principles apply.) The parallel filtering preferably consists of packet characteristics logic 22, packet type filters 26, and state rules filters 42. Packet characteristics logic 22 determines characteristics based on packet data (preferably in the form of 4-bit nibbles from PHY 14), whereas packet type filters 26 make filtering decisions generally based on packet type. State rules filters 42 perform rules-based filtering on several levels simultaneously. The results of filtering by packet type filters 26 and state rules filters 42 are combined by aggregator 24, which may be considered a type of logical operation of pass/fail signals (described in greater detail elsewhere herein). In preferred embodiments, if any one or more of the performed filtering rules indicates that the packet should be failed (or not allowed to pass as a valid packet), then the output of aggregator 24 is a fail; otherwise, the packet is allowed and the output of aggregator 24 is a pass. Thus, as packet data is being received and transmitted from PHY 14 to PHY 18 via repeater core 16, it is being evaluated in parallel via packet type filters 26 and state rules filters 42 (depending in part on packet characteristics determined by logic 22 from the data received from PHY 14). In accordance with the present invention, the results of filtering by packet type filters 26 and state rules filters 42 are provided to aggregator 24 by the time that the entire packet reaches repeater core 16, so that, based on the output of aggregator 24, the packet will either be allowed to pass as a valid packet or will be failed and junked as a suspect (or otherwise invalidated) packet.
Packet characteristics logic 22 receives packet data from PHY 14 and examines the packet data to determine characteristics, such as the packet type, datagram boundaries, packet start, packet end, data offset counts, protocols, flags, and receiving port. The packet type may include, for example, what are known in the art as IP, TCP, UDP, ARP, ICMP, or IPX/SPX. Such packet characteristics data is provided to packet type filters 26. Packet type filters 26 preferably make a decision about whether the packet should be passed or failed, with the result being transmitted to aggregator 24. In accordance with preferred embodiments, packet type filters 26 do not require the use of what may be considered an extensible rules system. The filters of packet type filters 26 preferably are expressed as fixed state machines or may be expressed using more flexible rules syntax. What is important is that packet type filtering is performed by filters 26 in the shortest time interval possible and in parallel with the packet data being received and transmitted to internal PHY 18, so that a pass/fail determination may be made prior to the time when the entire packet has been received by repeater core 16.
State rules filters 42 receive packet characteristics data from logic 22 and, based on this data as well as cached/stored connection and communication state information, executes a plurality of rules under the control of rules controller 28, preferably using a plurality of rules engines 36-1 to 36-N, so that a desired set of filtering decisions are promptly made and a pass/fail determination occurs before the entire packet has been received by repeater core 16. State rules filters 42 preserve a cache of information 30 about past network activity (such as IP addresses for established connections, port utilization, and the like), which is used to maintain network connection state information about which hosts have been exchanging packets and what types of packets they have exchanged, etc. Rules controller 28 preferably accesses rules map table 32 based on packet characteristics information, which returns rules dispatch information to rules controller 28. Thus, based on the connection state information stored in connection cache 30 and the characteristics of the packet being examined, rules controller 28 initiates filtering rules via a plurality of rules engines 36-1 to 36-N that simultaneously apply the desired set of filtering rules in parallel. (Preferably, N is determined by the number of rules that need to be performed in the available time and the speed of the particular logic that is used to implement state rules filters 42.)
As will be appreciated, while the packet pass/fail decision is being made in real time, and thus must be concluded by the time that the entire packet has been received, a large of number of filtering rules must be performed quickly and in parallel. Preferably, rules controller 28 utilizes a plurality of rules engines 36-1 to 36-N, which logically apply specific rules retrieved from corresponding storage areas 40-1 to 40-N. Rules controller 28, based on the connection state and packet characteristics, determines which rules should be run based on which information. The rules to be run are then allocated by rules controller 28 to the available rules engines 36-1 to 36-N. As each rules engine 36-1 to 36-N may be required to execute multiple rules in order to complete the filtering decision process in the required time, corresponding queues 34-1 to 34-N are preferably provided. Thus, rules controller 28 determines the list of rules that should be performed (again, depending on the stored connection state and packet characteristics data) and provides the list of rules (and accompanying information to carry out those rules) to the plurality of rules engines 36-1 to 36-N via queues 34-1 to 34-N. Rules engines 36-1 to 36-N, based on the information from the queues 34-1 to 34-N, look up specific rule information from storage areas 40-1 to 40-N, carry out the rules, and preferably return the results to rules controller 28. As the rules are essentially conditional logic statements that notify the data protection system how to react to a particular set of logical inputs, it has been determined that providing a plurality of rules engines may enable the necessary decision making process to quickly provide the outcome of the rules-based filtering by the time the entire packet has been received.
Still referring to
It should be appreciated that the data protection system must make a filtering determination before the current packet is completely transmitted. Since the networking standards impose strict timing thresholds on the transit delay of packets, filtering is performed in real time, in parallel and without buffering the packet. (The transit delay threshold is the time it takes to get from the transmitting station to the receiving station.) Given that a filtering decision must be made in real time (before the last bit is received and forwarded to the applicable interfaces), the filter rules are evaluated in parallel by rules engines that possess independent, direct access to the rules set collected in storage areas 40-1 and 40-N, which are preferably implemented as RAM tables. (In a preferred embodiment of the data protection system, the tables are implemented using on-chip, dual port RAM up to 4K in size. A programmable logic device, such as Xilinx Spartan II XC2S100 has 40K dual port synchronous block RAM. For example, an initial 110-bit segment of the rules controller RAM block may be a range table that delineates where each look up code begins and what the number of entries are.) Rules controller 28 dispatches the rules to each rules engine by placing a rules ID entry in a queue. Because each rules engine is assigned its own queue, a pipeline is created allowing the rules engine to continuously run and operate at maximum efficiency.
To operate efficiently the rules engines must also be capable of evaluating rules in any order. In accordance with the preferred embodiments, each rule has a priority and the highest priority result is accepted. Therefore, the rules must be evaluated in any order yet still obtain the same result, as if the rules were being evaluated serially from highest to lowest priority. This operation is accomplished in preferred embodiments by rules map table 32, which notifies rules controller 28 which rule is assigned to which rules engine. Thus, this decision is statically determined by the rules set and the number of rules engines. It should be noted that the rule set in general is greater than the number of rules engines.
As illustrated, after the packets are transmitted to repeater core 16, their characteristics are analyzed at step 44. Data packets generally consist of several layers of protocols that combine to make a protocol stack. Preferably, each layer of the stack is decoded and the information is passed to various filter blocks, as exemplified in steps 46, 48, 50 and 52. In accordance with the present invention, this filtering process is executed in parallel and in real time. In other embodiments, a variety of filter blocks or rules-based filters may be employed, incorporating parallel execution, real time filtering, etc., as may be necessary to complete the filtering decision in the required time.
Referring again to preferred embodiments illustrated in
In accordance with preferred embodiments of the present invention as illustrated in
As will be appreciated, block 60 in essence performs the repeater functionality of passing the incoming data to the non-active PHYs after reformatting the preamble. Block 60 also preferably receives “junk” or “pass” signals from the filtering components and a collision detection signal from PHY controller 56. In preferred embodiments, a “jam” signal is propagated to each PHY upon detection of a collision. A packet is invalidated for all PHYs that belong to a network category that receives a “junk” signal. (For example, if the packet is invalidated for internal networks, then the packet is invalidated for all internal network ports.) Preferably, block 60 also receives a single output signal from result aggregator 24 for each PHY category (i.e., internal or external). As will be explained in greater detail hereinafter, result aggregator 24 generates the signals provided to block 60 depending on “junk” or “pass” signals from each filter component.
In accordance with the present invention, the packet is also simultaneously routed through a plurality of filtering steps. In the exemplary illustration of Level 2 filters in
If it is a known packet type, then it is routed through additional filtering steps based on particular packet protocols. In the illustrated embodiment, at step 66, if the packet is an Address Resolution Protocol (ARP) type packet, then it is passed. At step 68, if the packet is a Reverse Address Resolution Protocol (RARP) type packet and is from external PHY 12 and the op code is 3, then it is junked; otherwise, it is passed as indicated at step 70. As is known in the art, RARP generally is a protocol used by diskless workstations to determine their address; in accordance with preferred embodiments, RARP responses are the only RARP packets allowed to enter internal networks from external hosts. At step 72, if the packet is an Internet Protocol (IP) type packet, is from the external PHY and has been broadcast, then it is junked. (For example, broadcast packets from the external network preferably are not allowed; a broadcast packet is determined by examining the IP address or the physical layer address). Otherwise, the process proceeds to step 74. Step 74 preferably examines the IP header, which contains a protocol fragment where an application can place handling options. Certain options (such as the illustrated list) may be considered to provide internal, potentially sensitive network information, and thus packets that contain these options preferably are not allowed into the internal network. At step 74, if a handling option of 7, 68, 131, or 137 is present, then the packet is junked; if these options are not present, then the process proceeds to filter IP packet step 76 (exemplary details of step 76 are explained in greater detail hereinafter). If the packet passes the filtering rules applied in filter IP packet step 76, then the packet is passed, as indicated by step 78. If the packet does not pass the filtering rules applied in filter IP packet step 76, then the packet is junked.
As illustrated in
At step 80, if the IP datagram type is unknown, then the fail signal is set, sending a signal to the result aggregator that the packet should be invalidated. At step 82, if the IP datagram type is Internet Group Management Protocol (IGMP), then the fail signal is set, preventing IGMP packets from passing. At step 84, if the type is Internet Control Message Protocol (ICMP) and the packet is from the external PHY, then the filtering proceeds to step 88. At step 84, if the type is ICMP and the packet is not from the external PHY, then the packet is passed as indicated by step 86. At step 88, if the type is ICMP, and the packet is from the external PHY and does not contain a fragment offset of 0, then the fail signal is set, preventing fragmented ICMP packets from passing, as indicated by step 90; otherwise, the filtering proceeds to step 92. At step 92, if the type is ICMP, the packet is from the external PHY and contains a fragment offset of 0, then the packet type is further evaluated for request and exchange data. This data preferably includes one of the following ICMP message types: 5 for redirect; 8 for echo request; 10 for router solicitation; 13 for timestamp request; 15 for information request; or 17 for address mask request. Accordingly, if the packet type satisfies the criteria for step 92, then the fail signal is set as indicated by step 96. Otherwise, the packet is allowed to pass, as indicated by step 94. As will be appreciated, the ICMP filtering branch serves to keep potentially harmful ICMP packets from entering from the external network. (The listed message types represent an exemplary set of ICMP packets that may expose the internal network topology to threats or cause routing table changes.)
If IP datagram characteristics indicate that the packet is a Transmission Control Protocol (TCP) or User Datagram Protocol (UDP) packet, then the filtering proceeds to step 98. At step 98, it is determined whether the packet is a fragment 0 packet. If it is not, then the packet is allowed to pass, as indicated by step 100. This filtering process follows the convention of filtering only the first fragments, as subsequent fragments will be discarded if the first one is not allowed to pass; in other words, the data protection system ignores all but the first packet of a TCP or UDP datagram. At step 104, if the packet is TCP or UDP and is a first fragment packet, then it is determined whether a proper protocol header is included in the fragment; if it is not, then the fail signal is set as indicated by step 102 (in the illustrated embodiment all TCP and UDP packets that have improper headers are junked). If the packet is TCP or UDP, is a first fragment, and a proper protocol header is included in the packet, then the filtering proceeds to step 106 (further exemplary details of which will be described in connection with
As indicated at step 114, if the internal port number is 68 and the external port number is 67, then the packet is passed, regardless of whether it originated on the internal network or the external network. As indicated at step 116, if the packet type is TCP, the server-mode is enabled (such as may be controlled by a toggle or other physical switch), the external PHY is active, and the internal port number is 80, then the packet is passed to the internal network(s). (The server mode is explained in greater detail in connection with
In preferred embodiments, if the completion signal is not generated by the time that the packet has been completely received, then the packet is junked. It should be noted that the use of such a completion signal and packet junking can be extended to the diagrams and description, etc. of other figures, such as
Referring now to
As previously described, Level 4 filtering is based on TCP and UDP packet characteristics, the determination of which is illustrated in
Mapping table 136 preferably is a ROM block that identifies the rules associated with each lookup code and the rules engine for which each rule is to be dispatched. The mapping data for the rules and rules engines are returned to rules dispatcher 134.
The identifiers of the rules to be run are dispatched by rules dispatcher 134 to the appropriate queues 138-1 to 138-N, which are preferably FIFO-type structures that hold the rule identifiers for corresponding rules engines 140-1 to 140-N. Queues 138-1 to 138-N not only enable rules dispatcher 134 to assign rules at maximum speed, but also allow each rules engine to retrieve rules as each one is evaluated. The rules engines 140-1 to 140-N are a plurality of filtering engines/logic:that use a rule table to read a definition specifying whether a rule applies to a packet and whether the packet passes or fails the rule test. Rules tables 142-1 to 142-N preferably are ROM blocks that contain a definition of a set of filtering rules that are controllably run by the rules engines 140-1 to 140-N. Rules tables 142-1 to 142-N may contain different rules as may be appropriate to provide all of the rules necessary to adequately filter packets within the timing constraints imposed by the real time filtering of the present invention, and the speed of the hardware used to implement the data protection system.
In addition, as illustrated in
In the illustrated embodiment, communication state information from rules engines 140-1 to 140-N may be provided to result aggregator 144, which in turn may store the communication state information to the communication state cache or storage area. Result signals, representing pass or fail of the packet based on the applied rules, also are provided to result aggregator 144. Result aggregator 144 combines the pass/fail results signals and provides a pass or junk signal or signals, which may be provided to the repeater core or to another result aggregator.
Stateful filters are implemented to handle communication channel interactions that span multiple transmissions between hosts. The interactions typically occur at the Application Layer of the protocol stack, where examples may include FTP, RealAudio, and DHCP. These interactions may also take place at lower levels in the protocol stack, such as ARP and ICMP request/response.
In this embodiment, stateful filters 154 use protocol front-end and protocol back-end logic, along with a plurality of state registers to implement state-dependent filters. Each protocol that requires stateful packet filtering preferably has protocol handlers in the form of front-end and back-end logic, which decide when to issue a pass signal for a packet or store the identifying characteristics of a bitstream for later reference. Front-end logic 160-1 to 160-N monitors the network traffic to identify when the current communication state needs to be stored, deleted or updated. Front-end logic 160-1 to 160-N informs a corresponding back-end logic 158-1 to 158-N that a register will be allocated for storage for a bitstream. All store and delete state register requests are sent to back-end logic 158-1 to 158-N so it may update its internal information. Register controller 155 controls the actual selection of registers in state registers 156 and informs the corresponding back-end logic 158-1 to 158-N. Back-end logic 158-1 to 158-N monitors which state registers are dedicated to its protocol and issues a pass signal for packets that match an existing bitstream, as indicated by the appropriate packet characteristics and a matching state register. It should be noted that in alternate embodiments, different organizations of the functions of the programmable logic may be implemented in accordance with the present invention, incorporating various types of protocol handlers and state registers, as may be necessary.
Register controller 155 consolidates multiple store and clear signals from the various front-end logic 160-1 to 160-N and directs them to the appropriate registers in state registers 156. Register controller 155 also informs the various back-end logic 158-1 to 158-N which registers of state registers 156 are to be used for storage. The registers of state registers 156, under control of register controller 155, store the communication state of a bitstream; for example, a particular register records information about the two communication ends of the bitstream and also monitors each network packet to see if it matches the stored end-point characteristics. State registers 156 then sets a signal when its state matches the current packet characteristics. A “garbage collection” function also is implemented (as further illustrated in
As is known in the art, many protocols provide a way of identifying the end of a communication session. Accordingly, in preferred embodiments the data protection system detects when a stateful stream ends and frees up the associated state registers. Since clients and servers do not always cleanly terminate a communication session, the system preferably implements session time-outs to free state registers after a period of bitstream activity and to prevent indefinite state register exhaustion. If the network experiences a high rate of bitstreams requiring stateful inspections, the system's resources, which are allocated to tracking application data, can become exhausted. In this case, the system preferably resorts to allowing network traffic based on a set of static rules to pass through the non-stateful rules designed specifically for each protocol. This stateful to non-stateful transition is called “stateful relaxation.” To maintain maximum security, a protocol handler that cannot gain access to an open state register will free up all of its state registers to help prevent other protocol handlers from entering into a relaxation state. The system will then wait for a state register to open, start a timer, and record protocol communication data in the state registers, while relying on the static rules. When the timer expires, the state filter will cease relying upon the static rules and approve packets solely on state register information.
The Internet connection, for example, via a cable modem, DSL router or other network interface, preferably is coupled with a physical cable to connector 168, which may be an RJ-45 connector. The signals received via connector 168 are coupled to and from PHY 170, which provides the physical interface for the data signals received from, or coupled to, the external network. Signals are coupled between PHY 170 and PLD 162, and signals are coupled between PLD 162 and PHY 172, which couples signals between connector 174 (which again may be an RJ-45 connector). The connection to the internal network may be made through connector 174.
In the preferred embodiment, PLD 162 implements the various levels of filtering as previously described. PLD 162 provides logic/hardware based, parallel filtering rules logic/engines, which make a decision about whether the packet should be allowed to pass or fail prior to the time that the packet is passed on by the repeater core portion of PLD 162 (as described elsewhere herein). The logic of PLD 162 to implement the filtering rules is programmed/loaded by controller 164, which may be a RISC CPU such as a MIPS, ARM, SuperH-type RISC microprocessor or the like. The PLD code preferably is stored in memory 166, which preferably is a re-programmable, non-volatile memory, such as FLASH or EEPROM. In this manner, the PLD code may be updated by reprogramming memory 166, and the updated PLD code may then be programmed/loaded in to PLD 162 under control of processor 164.
It should be noted that such visual feedback may be implemented in a variety of forms. In addition to multi-color or multiple LEDs or other lights sources or displays, a single LED could be provided, with the LED blinking at a rate that indicates the level of severity as predicted by the data protection system. For example, if no packets have been rejected, then the LED may be in an off or safe (e.g., green) state. If packets have been rejected but not on a continual or high rate basis, then the LED (e.g., red) may be controlled to blink on and off at a first, preferably lower speed rate. If packets are being rejected on a continual or high rate basis (or otherwise in a manner that that system believes is suspect), then the LED may be controlled to blink on and off at a second, preferably higher speed rate. Thus, the LED blink rate desirably may be controlled to blink at a rate that corresponds to the level of severity of the security threat that is determined by the data protection system. Optionally coupled with audio feedback, such visual indicators may provide the user with alarm and status information in a simple and intuitive manner.
As further illustrated in the preferred embodiments of
With reference to
In the illustrated embodiment, it is assumed that the user previously downloaded a system update or is downloading an update through a browser. The update program preferably breaks the update into 1K size packets and forwards them, using a limited broadcast destination address (for example, 255.255.255.255). The source and destination ports are set to a predetermined value, such as 1 (1-4 are currently unassigned according to RFC 1010), and an IP option is set in the IP header. The program data preferably is preceded by the system update header that has the following structure in the illustrated embodiment: ID (1)/count (1)/bit length (2). The numbers in parentheses represent the field size in bytes. The ID for the entire transaction remains unchanged, except for the count field increments for each packet. In a preferred embodiment, the data protection system may receive the packets in order and perform several checks, such as ensuring the ID and count fields are correct, verifying the UDP checksum, and storing the configuration data in non-volatile memory. Preferably, these checks may be controlled by controller 164. Thereafter, the updated PLD code may be loaded into the PLD, with the filtering operations being based on this updated code.
As a result of the parallel filter rules evaluation as previously described, packets do not need to be buffered, except, for example, to create octets that facilitate determining protocol elements. (As is known, data needs to be combined into 8-bit, 16-bit, or 32-bit words because header and packet data often exist in these sizes or straddle a 4-bit nibble boundary.) Instead of buffering each packet, the data protection system generates another distinct data packet or chunk. This process of packet generation occurs while a plurality of filtering rules are applied in real time and in parallel, producing improved data protection systems and methods.
In the preferred embodiment, on the internal network side of the U-shaped case, server mode button 200 is provided to allow the user to selectively enable filtering depending on whether the internal computer is allowed to operate in a server mode (thus, the state of server mode button 200 may be used to selectively control filtering decisions based on whether internal computers will be operating in a server mode, etc.). Server mode button 200 preferably includes server mode LED 202. When illuminated (e.g., green), server mode LED 202 indicates that the internal computers are enabled to operate in a server mode and the filtering decisions will be controlled accordingly. Server mode button 200 and server mode LED 202 are coupled to PLD 162, as described in
In a preferred embodiment, speaker 55 or some form of audio transducer may be coupled to alarm controller 53 to also indicate the presence or severity of attacks (as described in connection with
Adjacent to alert button 204 on the external network side of the case preferably is protection button 208, which is coupled to protection-on LED 212 and protection-off LED 214. When protection button 208 is set in the “on” position, protection-on LED 212 preferably illuminates red and the filtering system is enabled; when protection button 208 is set in the “off” position, protection-off LED 214 preferably illuminates yellow and the filtering system is disabled. As will be appreciated, the particular colors utilized are exemplary.
Still referring to
Adjacent to reset button 182 is update button 176, which is coupled to update-enabled LED 218 and update-disabled LED 220, as well as PLD 162 (as illustrated in
As further illustrated in
In an alternate embodiment, security levels switch 223 may be implemented to prevent stateful relaxation, in which a stateful to non-stateful transition may occur during state register exhaustion. As illustrated in
In other embodiments, different designs may be used in accordance with the present invention, incorporating various buttons, switches, LEDs, ports, cables, slots, connectors, plug-ins, speakers, and other audio transducers, which in turn may be embodied in a variety of external case shapes, as may be necessary. As will be appreciated, the filtering criteria may be dependent upon physical switch position, packet characteristics, clock time, and/or user-specified criteria, all of which may be entered through one or more physical input device(s). Such a physical input device, for example, may be comprised of one or more switches (such as a toggle switch, button switch, or multi-state switch), an audio input device, or display input device. The user-specified criteria may be transferred from the configuration software to the system using a network protocol, infrared port, or cable attachment.
As is known in the art, SYN flood is a common type of “Denial of Service” attack, in which a target host is flooded with TCP connection requests. In the process of exchanging data in a three-way handshake, source addresses and source TCP ports of various connection request packets are random or missing. In a three-way handshake, the system registers a request from an IP address, then sends a response to that address based on its source, and waits for the reply from that address.
As illustrated in
As illustrated in
In accordance with the present invention, SYN flood protection as described does not require either an IP or MAC address. The data protection system uses the destination MAC address as the source Ethernet address when framing the response packet that completes the TCP three-way handshake. In all cases, when forming the new packet, the source and destination header information is swapped, so that the source IP address and port become the destination IP address and port. It should be appreciated that SYN flood protection, as preferably implemented by the system, does not buffer the incoming packet, but builds the TCP response packet in real-time. The new TCP packet is placed in a queue for transmission at the earliest time possible based on the rules dictated by the link level protocol.
As illustrated in
Although the invention has been described in conjunction with specific preferred and other embodiments, it is evident that many substitutions, alternatives and variations will be apparent to those skilled in the art in light of the foregoing description. Accordingly, the invention is intended to embrace all of the alternatives and variations that fall within the spirit and scope of the appended claims. For example, it should be understood that, in accordance with the various alternative embodiments described herein, various systems, and uses and methods based on such systems, may be obtained. The various refinements and alternative and additional features also described may be combined to provide additional advantageous combinations and the like in accordance with the present invention. As will also be understood by those skilled in the art based on the foregoing description, various aspects of the preferred embodiments may be used in various subcombinations to achieve at least certain of the benefits and attributes described herein, and such subcombinations also are within the scope of the present invention. All such refinements, enhancements and further uses of the present invention are within the scope of the present invention.