US20160164807A1 - Systems and Methods For Efficient Handling of Data Traffic and Processing Within a Processing Device - Google Patents
Systems and Methods For Efficient Handling of Data Traffic and Processing Within a Processing Device Download PDFInfo
- Publication number
- US20160164807A1 US20160164807A1 US15/044,688 US201615044688A US2016164807A1 US 20160164807 A1 US20160164807 A1 US 20160164807A1 US 201615044688 A US201615044688 A US 201615044688A US 2016164807 A1 US2016164807 A1 US 2016164807A1
- Authority
- US
- United States
- Prior art keywords
- packet
- data
- agent
- interconnect
- adapter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/70—Virtual switches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F13/00—Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
- G06F13/38—Information transfer, e.g. on bus
- G06F13/40—Bus structure
- G06F13/4004—Coupling between buses
- G06F13/4027—Coupling between buses using bus bridges
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/40—Constructional details, e.g. power supply, mechanical construction or backplane
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Computer Hardware Design (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention provides an improved platform hub that aims to, in some embodiments, optimize system resources to improve system performance and/or reduce consumption of power.
Description
- This application is a continuation of U.S. application Ser. No. 13/474,932, filed on May 18, 2012 (published as U.S. 2013-0016729), which is a continuation of U.S. application Ser. No. 12/875,343, filed Sep. 3, 2010 (now U.S. Pat. No. 8,209,457), which is a continuation of U.S. application Ser. No. 11/776,285, filed Jul. 11, 2007 (now U.S. Pat. No. 7,793,032). The above identified applications and publications are incorporated by reference.
- 1. Field of the Invention
- The present invention relates generally to processing devices, and, more specifically to systems and methods for improving the handling of the movement of data and processing of data within a processing device.
- 2. Discussion of the Background
-
FIG. 1 illustrates aconventional processing device 100.Processing device 100 includes ahousing 102 that houses one ormore agents 110 and aplatform hub 120. For example,housing 102 may house the following agents: zero or more acceleration agents 110 a(1)-110 a(N), zero ormore processing agents 110 b(1)-110 b(N), zero ormore communication agents 110 c(1)-110 c(N), zero ormore storage agents 110 d, zero ormore legacy agents 110 e, zero or moreexternal memory agents 110 f, etc. Acommunication agent 110 c may include a physical connection to a communication channel (e.g., a fiber cable or Ethernet cable). Theplatform hub 120 is sometimes referred to as a “south bridge (SB)” or “north bridge (NB).” Aprocessing agent 110 b may be an x86 microprocessor from Intel, or may be processing device sold by AMD or other source of processing devices, and theplatform hub 120 may be contained in a chip that sits on the same circuit board as aprocessing agent 110 b. -
FIG. 2A illustrates oneexample processing agent 110 b. As shown inFIG. 2 , aprocessing agent 110 b may include (a) ahost 202 that includes one or more central processing units (CPU) and (2) one or more memory banks coupled tohost 202. In the example shown, there are two memory banks, one that is positioned to right of thehost 202 and one that is positioned to the left of thehost 202.FIG. 2B attempts to illustrate anotherexample processing agent 110 b. As illustrated inFIG. 2B , aprocessing agent 110 b can also be a multi-processor cluster (e.g. AMD Opteron™ x86 System). In such a case, the platform hub can be connected to one or a subset of processors in the cluster through a SB/NB Link, but can, in most cases, interact with all of the processors or memories in the cluster since the cluster is interconnected. - Platform Hub 120 is configured to enable the
agents 110 to communicate with aprocessing agent 110 b, but is not configured to enable the “non-processing agents” (e.g., agents 110 a,c,d,e,f) to communicate directly with each other, but this is not the only disadvantage ofplatform hub 120. - Thus, in the
conventional processing device 100, all data must flow through aprocessing agent 110 b. That is, for example, if data output from acommunication agent 110 c is ultimately destined for astorage agent 110 d, the data output fromcommunication agent 110 c is received byplatform hub 120, which then provides the data to aprocessing agent 110 b such that the data is stored in a memory unit of the processing agent. After the data is stored in the memory unit, the data is then received from the memory unit and provided toplatform hub 120, which then provides the data tostorage agent 110 d. Each data movement transaction consumes system resources due to the data handling and consumes system power due to the data passing process. - Some embodiments of the present invention aim to improve the data handling and/or passing process so as to reduce the amount of system resources and/or power that is consumed.
- In one aspect, the invention provides a processing device. In one embodiment, the processing device includes: a housing; a platform hub housed in the housing; and an agent housed in the housing and connected to the platform hub, wherein the platform hub comprises: an interconnect; and a classification adapter unit, wherein the classification adapter unit is connected between the agent and the interconnect, and the classification adapter unit is configured to interface with the agent such that the classification adapter unit may obtain data from the agent and may provide data to the agent. The data may be a protocol packet (e.g., a TCP/IP packet).
- In another aspect, the invention provides a chip for use in a processing device. In one embodiment, the chip includes: an interconnect; a first classification adapter unit circuit directly connected to the interconnect; and a second classification adapter unit circuit directly connected to the interconnect, wherein the interconnect is configured to (a) enable the first classification adapter unit circuit to send data to and receive data from the second classification adapter unit circuit and (b) enable the second classification adapter unit circuit to send data to and receive data from the first classification adapter unit circuit, and the first classification adapter unit circuit is operable to: (1) receive a block of data from an agent, (2) add a directive to at least a portion of the data block, thereby creating a data container, and (3) transmit the data container to the second classification adapter unit circuit by providing the data container to the interconnect.
- In another aspect, the invention provides a method. In one embodiment, the method includes: receiving, at a first classification adapter unit, a data block from an agent; creating a directive in response to receiving the data block; adding the directive to the data block, thereby creating a data container comprising the directive and the block of data; providing, from the first classification adapter unit and to an interconnect, the data container created by the first classification adapter unit; receiving the data container at the interconnect, wherein the interconnect provides the data container to a second classification adapter unit; and receiving the data container at the second classification adapter unit, wherein the second classification adapter unit performs an action based, at least in part, on information included in the data container, and the first classification adapter unit, the second classification adapter unit, and the interconnect are built upon a chip and the chip is directly connected to a circuit board.
- In some embodiments, the data block is a protocol packet and the step of creating the directive comprises: examining the protocol packet and creating the directive based, at least in part, on information contained in the protocol packet. The step of examining the protocol packet may include (a) examining a header portion of the protocol packet and/or (b) performing a deep packet inspection (i.e., analyzing the payload of the packet in addition to or instead of the header of the packet) (e.g., searching for a pattern or for certain data inside the packet payload).
- The above and other embodiments of the present invention are described below with reference to the accompanying drawings.
- The accompanying drawings, which are incorporated herein and form part of the specification, illustrate various embodiments of the present invention. In the drawings, like reference numbers indicate identical or functionally similar elements.
-
FIG. 1 . illustrates a conventional processing device. -
FIG. 2A illustrates a conventional processing agent. -
FIG. 2B illustrates another conventional processing agent. -
FIG. 3 illustrates a platform hub according to some embodiments of the invention. -
FIG. 4A illustrates an example data container. -
FIG. 4B illustrates an example protocol packet. -
FIG. 5A illustrates an example instruction structure. -
FIG. 5B illustrates two way in which an instruction may be accessed. -
FIG. 6 is a functional block diagram of a classification adapter unit according to one embodiment. -
FIG. 7 illustrates an example data flow. -
FIG. 8 is a flow chart illustrating a process according to an embodiment of the invention. -
FIG. 9 illustrates an example data container. -
FIG. 10 illustrates an example data flow. -
FIG. 11 is a flow chart illustrating a process according to an embodiment of the invention. -
FIG. 12 illustrates an example data flow. -
FIG. 13 is a flow chart illustrating a process according to an embodiment of the invention. -
FIG. 14 illustrates an example directive. -
FIG. 15 illustrates an example data flow. -
FIG. 16 is a flow chart illustrating a process according to an embodiment of the invention. -
FIG. 17 illustrates an example data flow. -
FIG. 18 is a flow chart illustrating a process according to an embodiment of the invention. -
FIG. 19 illustrates an example directive. -
FIGS. 20A ,B illustrate an example data flow. -
FIGS. 21A ,B is a flow chart illustrating a process according to an embodiment of the invention. -
FIG. 22 illustrates an example classification adapter unit. -
FIGS. 23-26 illustrate various other embodiments of the invention. - As used herein, the words “a” and “an” mean “one or more.”
- In one embodiment, the present invention provides an
improved platform hub 320, which may be contained in a chip, and which chip may be attached to the same circuit board as aprocessing agent 110 b. -
Platform hub 320, in contrast toconventional platform hub 102, may, in some embodiments, be configured to, among other things, enable the non-processing agents to communicate with each other without having to involve a processing agent. Other features that may be implemented byplatform hub 320 are described below.Platform hub 320 may replace an existingplatform hub 120 or may be used in conjunction with an existingplatform hub 120. - As illustrated in
FIG. 3 ,platform hub 320 may include one or more classification adapter units 310 (or “adapters” for short) coupled to aninterconnect 330 for receiving data from and transmitting data to aclassification adapter unit 310.Interconnect 330 may include a router, a switching circuit (e.g., a crossbar switch or other switching circuit), etc. - In the illustrated embodiment, each
classification adapter unit 310 is connected between anagent 110 andinterconnect 330. Eachadapter 310 may be configured to be able to obtain data directly from theagent 110 to which it is coupled. This data is referred to as a “data payload.” This data payload may be a protocol packet (e.g., a TCP packet), a block of information that was created by the agent in order to encapsulate a transaction of some sort, or some other block of data. As illustrated, some agents may be included withinhub 320, while others exist outside ofhub 320. Additionally, an agent may be included in anadapter unit 310. For example, an agent, such as a central processing unit, my be included in anadapter unit 310. - In some embodiments, an
adapter 310 may be configured to (a) obtain from an agent 110 a data payload (e.g., a block of data) (b) examine at least some of the data contained in the data payload and (c) take an action based on the examined data. The action taken may include one or more of the following: (1) communicating information to anotheradapter 310, (2) communicating information to aprocessing agent 110 b, (3) modifying the data payload, (4) discarding the data payload, etc. - For example, in some embodiments, a communication adapter (i.e., an
adapter 310 connected to acommunication agent 110 c) may be configured to (a) receive from a communication agent a data payload (e.g., a protocol packet such as a TCP/IP packet), (b) examine a header contained in the data payload (e.g., a TCP/IP header), and (c) take some action (e.g., drop the TCP/IP packet) based on information contained in the header of the packet and/or information configured in the adapter 310 (e.g., a list of known viruses, a list of the last 50 flows that exited in the system). - In some embodiments, one
adapter 310 may have a different structure and may perform different functions from anotheradapter 310. For example, a classification adapter 31 that is coupled to an acceleration agent 110 a may have different functionality and/or a different structure than aclassification adapter 310 that is connected to aprocessing agent 110 b. As another example, aclassification adapter 310 that is coupled to acommunication agent 110 c(1) may have different functionality and/or a different structure than aclassification adapter 310 that is coupled to acommunication agent 110 c(2). - More specifically, in some embodiments, an
adapter 310 may be specially designed to match the nature of the agent to which it is connected. This makes the physical design of each of theadapters 310 optimal for the work that needs to be done on data payloads coming from that agent. For example, an adapter servingcommunication agent 110 c(1) may be specially designed to match the nature of the port to which it is connected throughagent 110 c(1). Accordingly, the design of such an adapter may be optimized for the types of protocol packets that are received by the port to which it is coupled. The distribution to small, different, dedicated adapter enables a hardware implementation that can satisfy the full bandwidth of the port. - In some embodiments, each
adapter 310 may implement a directive driven architecture. For example, eachadapter 310 may be configured such that it is able to add a “directive” to a data payload received from anagent 110, thereby creating a data container. Thus, this new data structure (the data container) contains information (i.e., the directive) “glued” to the data payload, preferably, at an accessible place (e.g., at the head or tail of the data payload). The encapsulation of a directive in—band with a data payload may allow for efficient use of resources via, for example, effective routing, delivering statistic platform information, and actions to be taken on the data payload. - The directive may include any combination of the following: routing information, quality of service information, vlan-tag insertion commands, information characterizing the data payload (e.g., “this data payload is a media container”), path related information (e.g., information regarding the path the data payload traversed), platform information (e.g., congestion, device status), etc.
- Thus, the data payload-directive coupling gives the ability to pass control information in-band with the data payload, instead of passing it out-of-band in a different control path. The in-band control passing method enables the
platform hub 320 to be scalable and be able to linkother platform hubs 320. - The directive added to a data payload by an
adapter 310, can be used byinterconnect 330 to, among other things, determine the adapter to which the data container should be directed. - The directive can also be used by the
adapter 310 that receives the data container from theinterconnect 330. For example, theadapter 310 that receives the data container frominterconnect 330 can examine the directive to determine whether an action should be performed. This adapter, based on the directive, may take any combination of the following actions: a quality of service action, a routing action, modify the data in the data payload, add a CRC to the data container, add virtual LAN tags to the data, decrypt the data payload, collect/gather further platform statistics (e.g., congestion information, usage statistics) to be used by other objects, etc. -
FIG. 4A illustrates adata container 400 according to some embodiments of the invention. As shown inFIG. 4A ,data container 400 includes adata payload 402 that was produced by anagent 110 and a directive 404 that was added to thepayload 402 by theadapter 310 connected to theagent 110, whichadapter 310 received thepayload 402 from theagent 110. In the example shown inFIG. 4A the directive 404 is attached to the head of thedata payload 402, but it may also be attached at other locations, such as the tail. - The directive 404 may be a simple bit vector. But, in some embodiments, it is a program-oriented structure because such a structure may be more efficient, especially in an in-line distributed classification architecture. In the embodiments in which the directive has a program oriented structure, the directive may include a set of instructions that can be random-accessed and/or stack-like accessed.
-
FIG. 4B illustrates anexample data payload 402. In the example shown inFIG. 4B ,data payload 400 is aprotocol packet 450, whichpacket 450 includes apacket header portion 451 and a protocol packetdata payload portion 452. - The set of instructions attached to a data payload by an
adapter 310 may form a sequence of operations to be performed on the data payload.FIG. 5A illustrates one example of the structure of such functions. As illustrated inFIG. 5A , an instruction may include three parts: (1) a command operation code (“command Opcode”) 501; (2)operands 502; and (3)immediate information 503. - In some embodiments, the directive may include a list of instructions. The instructions in the list may be accessed randomly or stack based, as illustrated in
FIG. 5B . There may be advantages to stack based access of the instructions. For example, eachadapter 310 that receives a data container can pop the top instruction in an O(1) of time. It then can perform the action required by the popped instruction. It can also use the instruction in order to pass an instruction to an adapter through a dedicated control path. When the packet returns from the adapter, the adapter can add an operation to the top of the stack by pushing it into the directive in an O(1) amortized time. Adding an instruction to the top of the stack enables each adapter to decide to spread its task over a set of other adapters without affecting the rest of the exaction path. The directive stack-like implementation resembles a CPU software stack in most systems. It enables performing sub-routines without harming the upper level execution flow. Random Access to the instruction set enables out-of-order execution and insertion. - Referring now to
FIG. 6 ,FIG. 6 illustrates a functional block diagram of an adapter according to some embodiments of the invention. As illustrated inFIG. 6 , the adapter may include two distinct data paths, a transmit (TX) data path and a receive (RX) data path. - As further illustrated, the TX data path may include an
agent interface 604 for interfacing theadapter 310 with anagent 110, a direct memory access (DMA)engine 603, an action engine (AE)circuit 602, and aclassifier circuit 601. In the embodiment shown,classifier 601 is connected directly betweeninterconnect 330 andAE 602;AE 602 is connected directly betweenclassifier 601 andDMA 603;DMA 603 is connected directly betweenAE 602 andinterface 604; andinterface 604 is connected directly betweenDMA 603 andagent 110. - The RX data path includes an
interface 614 for interfacing withagent 110, aDMA engine 613, aclassifier 612, and anAE 611. In the embodiment shown,AE 611 is connected directly betweeninterconnect 330 andclassifier 601;classifier 612 is connected directly betweenAE 611 andDMA 613;DMA 613 is connected directly betweenclassifier 612 andinterface 614; andinterface 614 is connected directly betweenDMA 613 andagent 110. - Although a specific arrangement is shown in
FIG. 6 , this is for illustration and should not limit the scope of the invention as it is contemplated that one or more of the illustrated components are optional. For example, in some embodiments the DMA is not used. - Agent interfaces 614 and 604 may be responsible for translation between an agent and the platform hub. This includes handshakes between the platform hub and the agent. For example,
interface 614 may be configured to receive data from a particular type of agent and may format the data according to a predetermined format prior to providing the data toDMA 613. Because eachagent 110 may provide data to aninterface 614 in a different format, eachagent interface 614 is designed specifically for the agent to which it interfaces. Accordingly, in some embodiments, agent interfaces 604 and 614 are modular such that they can be easily replaced. Agent interfaces 604, 614 may also encapsulate a physical link unit that is link-dependent (i.e., PCIe, HyperTransport, etc.). - In some embodiments,
DMAs DMA 603 may be operable to write data to a certain memory location and then send a command to an agent, which command may cause an interrupt to occur which causes the agent to read the data from the certain memory location. As another example, an agent may send a command to aDMA 613 that causes the DMA to read the data stored in a predefined memory location. More specifically, for example, anacceleration agent 110 that is configured to encrypt data may send a command to a DMA of anadapter unit 310 immediately after the acceleration agent encrypts the data and stores the data in a certain memory, thereby causing the DMA to retrieve the encrypted data from the memory. After retrieving the encrypted data, the DMA may provide the data to aclassifier 612 for further processing as described herein. - As described above, a DMA can interact with an agent in various manners dependant on the nature of the agent. In general there could be 3 main logical connections between an agent and a DMA: (1) data buffers, (2) data descriptors, and (3) control. All connections may be used both ways, and in some agents only a subset or a variation of these exist. The control path between the DMA and the agent is used to pass commands, instructions and configurations between the two. Commands from the DMA to the agent can also trigger specific sub-units inside an agent. This is very useful to instruct the required specific sub-unit to perform the task at hand. Using the command path, the DMA can control not only the agent as a whole but also as a cluster of sub-units. For example, when a packet is sent to a processing agent that includes several processors, the packet may be sent to a memory of one of the processors. In such case, the DMA can send a control message which triggers an interrupt on this specific processor and indicates that it has a packet waiting for it. This provides a benefit by utilizing system resources in an optimized manner. A different example can be in an Acceleration Agent. In such a case, the DMA performs a write of a packet to the agent memory space using the data path. When it is finished writing the packet, it send the agent a control message that instructs the agent to start the designated operation on the packet. When the agent is done, the agent writes a control message to the DMA that the operation is finished and the DMA reads the packet back from the acceleration agent. In most of the cases in which there is no (or little use) of a command path, the agent may poll a data/descriptor memory area in order to act on each packet.
-
Classifier 612 may be configured to parse the data it receives fromDMA 613, extract relevant fields from the data, and take an action (e.g., create or select a command) based, at least in part, on the relevant data fields. Ifclassifier 612 selects or creates a command, the command may be passed toAE 611 and may instructAE 611 to take a specific action. In addition to providing commands toAE 611,classifier 612 may also provide toAE 611 data it received fromDMA 613. -
Classifier 601 may be configured to receive data containers frominterconnect 330 perform data classification based on the extraction of instructions included in the directive portion of a data container received frominterconnect 330. - A
classifier classifier 612 may include a parser and an identifier. The parser may be configured to classify a packet (IP/TCP/UDP etc . . . ) and extract relevant fields from the packet, while the identifier may check the fields against relevant rules. Thus, in some embodiments, aclassifier 612 may include a rules engine that implements a set of rules on a set of fields. In some embodiments, the output of a classifier is a list of commands accompanied by extra information, if necessary, that could be of use by the AEs or one of the agents. - The
AEs - In the example illustrated in
FIG. 6 , theclassifiers adapter 310 can have the ability to directly communicate with any other component ofadapter 310. -
Interconnect 330 enables connectivity between the different adapters. In some embodiments,interconnect 330 is a non-blocking switch that uses the data container as its only possible data structure. The interconnect itself is a kind of an engine as well, being capable to perform a “MOVE” instruction. Since the structure of the directive is a stacked list of instruction, a classifier can decide a multiple hops action. This will be done by inserting several “MOVE” instruction one after the other. - Referring now to the RX path of
adapter 310,DMA 613 may receive a data payload fromagent 110 viainterface 614 and provides the data payload toclassifier 612. In response to receiving the data payload fromDMA 613,classifier 612 may examine the data payload and then pass toAE 611 the data payload along with classification information that is used byAE 611 to create a directive to add to the data payload. This classification information may be passed out-of-band toAE 611. The classification information may depend on the contents of the payload. For example,classifier 612 may examine one or more fields of the data payload, and, depending on the data in those fields, select certain commands to include in the classification information sent toAE 611. - In response to receiving from
classifier 612 the data payload and classification information,AE 611 may be configured to create a directive and add the directive to the data payload, thereby creating a data container. The directive created byAE 611 may depend on the classification information it received fromclassifier 612. After creating the data container,AE 611 may provide the data container to interconnect 330, which may be configured to route the data container to anadapter 310 specified in the directive. - Referring now to the TX path of
adapter 310, as discussed above,classifier 601 may receive data containers 600 frominterconnect 330 and may add and/or remove information from the container'sdirective 604. Afterclassifier 601 is finished processing a received data container 600, it may pass the data container 600 toAE 602.AE 602 may perform steps depending on the information contained in the data container's directive and/or depending on commands received fromclassifer 601. For example,AE 602 may pass the container's data payload toDMA 603. AE 620 may also send out-of-band control information toDMA 603.DMA 603 receives data payloads fromAE 602 and provides the received data payloads toagent 110 viainterface 604. - Referring now to
FIG. 22 ,FIG. 22 illustrates an example RX path of anadapter 310 that is connected to aprocessing agent 110 b. In the example shown, we shall assume thatDMA 613 obtains a packet descriptor from a memory unit ofagent 110 b and obtains a protocol packet from a memory unit ofagent 110 b, wherein the protocol packet is associated with the packet descriptor. The packet descriptor may contain a set of fields. For example, the packet descriptor may includes the following data fields: packet size field that identifies the size of the protocol packet, a destination port field that identifies a destination for the protocol packet, a quality of service field that may identifier a packet queue; a CPU identifier; etc. - After the
DMA 613 obtains the packet descriptor and protocol packet, it may pass the packet descriptor to aparser 2201 ofclassifier 612 using a first bus 2210 and may pass the protocol packet to theparser 2201 using a second bus 2211. -
Parser 2201 may be configured to extract fields from the packet descriptor and may be configured to extract fields from the header of the protocol packets. The Extracted fields may be provided to anidentifier 2202 and the protocol packet may be provided to theaction engine 611. -
Identifier 2202 may be configured to compare a field received fromparser 2201 to configuration data to determine an action that should be taken. Based on the determined action, theidentifier 2202 may send a command to theaction engine 611.Identifier 2202 may also provide to theaction engine 611 the extracted fields. - For example, if
identifier 2202 determines that the destination port of the protocol packet isport # 2, thenidentifier 2202 may determine that it should send a “merge” command to theaction engine 611. As discussed above,action engine 611 may perform the action and may create a data container, which is then provided to interconnect 330. - Referring now to
FIGS. 7 and 8 ,FIG. 7 shows an example flow of data throughplatform hub 320 andFIG. 8 is a flow chart describing the steps of the data flow. The example data flow begins instep 801, wherein thecommunication agent 110 c(1) receives a data payload to be delivered toprocessing agent 110 b(1). For example,communication agent 110 c(1) may receive a TCP/IP packet from a network and this TCP/IP packet may need to be delivered toprocessing agent 110 b(1) so that the packet can be processed. - In
step 802, the adapter that servesagent 110 c(1) (i.e.,adapter 310 c(1)) receives the payload fromagent 110 c. Instep 804, the adapter parses the data payload (e.g., the adapter examines the header of the TCP/IP packet) to determine the stream or connection to which the packet belongs (in this example, we shall assume the packet belongs to stream N). One objective ofadapter 310 c may be to perform such classification operation in a persistent manner, hence sending packets of the same logical and operational characteristics (need similar operations to be performed upon them) to thesame processing agent 110 b. - In
step 805, the adapter retrieves and/or generates certain classification information in response to determining that the packet belongs to stream number N. That is, the classification information that is generated/received depends, at least in part, on the fact that the packet belongs to stream N. - In
step 806, the adapter performs the following steps: (1) creates a directive that includes all the relevant data that was gathered/created instep 805, if any, along with a list of instructions to be performed on the data payload, (2) adds the directive to the head of the data payload, thereby creating a data container, and (3) passes the data container to interconnect 330.FIG. 9 illustrates an example directive that may be created instep 806. The example directive includes twomove instructions extra information record 903. - In
step 807,interconnect 330 receives the data container fromadapter 310 c(1) and removesinstruction 901 from the directive and routes the data container to the port identified in move command 901 (i.e., port #3), which is the port to which the adapter forprocessing agent 110 b(1) is attached. - In
step 808,adapter 310 b(1) receives the data container frominterconnect 330, pops the second move instruction (i.e., instruction 902) from the directive, which instruction indicates the ultimate destination for the data payload (i.e.,host memory # 1 in this example). - In
step 810, theadapter 310 b(1) causes the data payload to be stored inhost memory # 1 and also causes at least some of the information contained inrecord 903 to be stored in a predefined descriptor storage area, which may be inhost memory # 1. - Referring now to
FIGS. 10 and 11 ,FIG. 10 shows an example implementation of the data flow shown inFIG. 7 andFIG. 11 is a flow chart describing the steps of the example implementation. - The data flow begins in
step 1102, wherein thecommunication agent 110 c(1) receives a data payload to be delivered toprocessing agent 110 b(1). For example,communication agent 110 c(1) may receive a TCP/IP packet from a network and this TCP/IP packet may need to be delivered toprocessing agent 110 b(1) so that the packet can be processed. - In
step 1104, the interface (e.g. an Ethernet Mac interface) receives the payload fromagent 110 c and translates the information from 1st layers. - In
step 1106, the payload is passed to theDMA 613, which in this simple scenario doesn't need to perform any batch operation (in a more complicated example it could decide for instance to perform a back-up write operation to a storage device while passing the data to the classifier), andDMA 613 passes the data payload toclassifier 612. - In
step 1108, theclassifier 612 parses the data payload to determine the stream or connection to which the packet belongs (in this example, we shall assume the packet belongs to stream number N). - In
step 1110, theclassifier 612 retrieves and/or generates certain classification information based, at least in part, on the fact that the packet belongs to stream number N. It then passes, out-of-band, toAE 611 the classification information and also passes the data payload toAE 611. - In
step 1112,AE 611 receives the classification information, creates a directive that includes all the relevant data that was gathered through out the classification phase along with a list of instructions to be performed on the data payload, adds the directive to the head of the data payload, thereby creating a data container, and passes the data container to interconnect 330. - In
step 1114,interconnect 330 removesinstruction 901 from the directive and routes the data container to the port identified in move command 901 (i.e., port #3), which is the port to which the adapter forhost 202 is attached. - In
step 1116,classifier 601 receives the data container frominterconnect 330, pops the second move instruction (i.e., instruction 902) from the directive, which instruction indicates the ultimate destination for the data payload (i.e.,host memory # 1 in this example), passes the data container toAE 602, and passes, out-of-band, to the AE 602 a command to send the data payload to the memory identified in the move instruction (i.e., host memory #1). - In
step 1118,AE 602 popsrecord 903 and passes, out-of-band, to theDMA 603 the extra information contained therein. Instep 1120,DMA 603 performs a store request to hostmemory # 1 of all packet data. The extra information is sent to a predefined descriptor area also inhost memory # 1. Instep 1122,interface 604 translates the request to the host specific bus. Instep 1124, data payload is stored inhost memory # 1. - The scalable structure of
platform hub 320 enables fast, efficient and scalable control over the system platform agents. By using the directive structure, control signals in the system can be passed as in-band messages (in fact, signals can be even sent as piggy-backed information upon existing traffic in the opposite direction). An example of a control messages is an instruction fromhost 202 to one ormore classifiers 612 to watch out for a new virus. In such scenario, host 202 sends a message to some or alladapters 310, in a request to update their tables according to the new classification scheme. Host 202 can update theadapters 310 by performing a memory-mapped-io-write to the adapters or by sending a control packet with a predefined packet format. - Referring now to
FIGS. 12 and 13 ,FIG. 12 shows another example flow of data throughplatform hub 320 andFIG. 13 is a flow chart describing the steps of data flow. - Referring to
FIG. 13 , instep 1302, host 202 outputs data (1), which data is received byadapter 310 b(1). In this example, the data includes an adapter configuration command or “management packet” that should be sent to other adapters (e.g.,adapters 310 a,c,d. - In
step 1304,adapter 310 b(1) examines the received data and determines that it includes an adapter configuration command. Because the data includes an adapter configuration command,adapter 310 b(1) creates a certain directive (2) and provides the directive to interconnect 330 (step 1306). In some embodiments, the directive is provided to interconnect 330 by piggybacking on existing packet streams. That is, in some embodiments, the directive is added to a data payload to create a data container, and the data container is sent to interconnect 330. -
FIG. 14 illustrates an example directive that may be created instep 1304. As illustrated inFIG. 14 , the directive created instep 1304 may include two instructions: (1) an instruction for the interconnect 330 (instruction 1401, which in this example is a MOVE instruction) and (2) another instruction for the other adapters (instruction 1402). - In
step 1308,interconnect 330 receives the directive, removes (pops) the MOVE instruction from the directive, which move instruction instructsinterconnect 330 to provide the remaining instruction toports - In
step 1310, the adapters connected toports adapters instruction 1402 from the interconnect and update their configuration information (e.g., configuration tables) accordingly. In this example, the instruction tells the adapters to drop a packet if the packet meets a certain criteria specified ininstruction 1402. - In
step 1312, anew packet 3 that meets the specified criteria is received at thecommunication agent 110 c(1), which outputs the packet such that it is received byadapter 310 c(1). When this packet reachesadapter 310 c(1), the packet and the adapter's configuration information are examined by the adapter to determine whether the adapter needs to take any action with respect to the packet (step 1314). Because, in this example, the packet meets the criteria specified in instruction 1402 (e.g., the packet is a malicious packet), the adapter performs the action specified in instruction 1402 (i.e., the adapter drops the packet) (step 1316). - Referring now to
FIGS. 15 and 16 ,FIG. 15 shows an example implementation of the data flow shown inFIG. 12 andFIG. 16 is a flow chart describing the steps of the example implementation. - In
step 1601, host 202 performs a write to theplatform hub 320. Instep 1602, the agent interface connected to host 202 translates the register write into a predefined-format (e.g., a fixed-length packet) and provides the packet to the DMA to which it is connected. Instep 1603, the DMA passes the packet to the classifier to which the DMA is connected. - In
step 1604, the classifier examines the packet and identifies that the packet as a “management packet”. The classifier then issues a request to the AE to create a certain directive and provide the directive to interconnect 330. Instep 1605, the AE creates the required directive and provides it to interconnect 330. - In
step 1606,interconnect 330 receives the directive, removes (pops) the MOVE instruction from the directive, which move instruction instructsinterconnect 330 to provide the remaining instruction toports classifier 601 in each of the other three adapters. - In
step 1607, eachclassifier 601 in the other adapters receives the instruction 1202 from the interconnect and identifies that the instruction should be sent to its neighboringclassifier 612. Eachclassifier 601 then passes the instruction to its neighboringclassifier 612. Instep 1608, eachclassifier 612 receivesinstruction 1402 and updates its tables accordingly. In this example, the instruction tellsclassifier 612 to drop a packet if the packet meets a certain criteria specified ininstruction 1402. - In
step 1609, a new packet that meets the specified criteria is received at the communication channel. When this packet reaches theclassifier 612 ofadapter 310 c(1), the packet is examined by the classifier and the classifier issues a drop command to the AE because the packet meets the specified criteria (i.e., it is a malicious packet) (step 1610). The AE then drops the malicious packet (step 1611). - Referring now to
FIGS. 17 and 18 ,FIG. 17 show another example flow of data throughplatform hub 320 andFIG. 18 is a flow chart describing the steps of the data flow. - Referring now to
FIG. 18 , the data flow may begin instep 1801, whereincommunication agent 110 c(1) receives a protocol packet to be delivered toprocessing agent 110 b(1). Instep 1802,adapter 310 c(1) receives the packet fromcommunication agent 110 c(1). - In
step 1803,adapter 310 c(1) parses the protocol packet and determines the stream to which the packet belongs (e.g., it may identify that the packet as belonging to a certain stream (e.g., stream #N)) and determines, based on the determined stream, whether the protocol packet should be “split.” A split operation means that, for example, certain data and/or only a portion of the packet (e.g., the first X number of bytes of the packet, where X is greater than or equal to 0) need be provided toprocessing agent 110 b(1), while a copy of the entire packet should be copied tostorage agent 110 d (which may be referred to as “platform memory”). For this example, we shall assume the packet should be split. - In
step 1804, the adapter builds a vector of fields. This field-vector may include: fields that were extracted from the packet header (e.g., source/destination addresses Ethernet/IP etc.), fields extracted from the packet application data, data resulting from a certain operation (e.g. tupliz-hash calculation on the packet data/header-fields) and/or data retrieved from storage (e.g., from stored configuration information). - In
step 1805,adapter 310 c(1) creates a first data container that includes a data payload and a directive. The data payload includes the first X bytes of the protocol packet (X >=0). The directive includes: (1) routing information that instructsinterconnect 330 to provide the data container toadapter 310 b, (2) the field-vector and (3) a unique identifier that represents the protocol packet and is used, in a later phase, to retrieve the protocol packet. This data container is then sent to theprocessing agent 110 b(1). - In
step 1806,adapter 310 c(1) creates a second data container that includes a data payload and a directive. The data payload includes the entire application data portion of the protocol packet (it may also include the headers). The directive includes: (1) routing information that instructs interconnect to provide the data container toadapter 310 d and (2) the unique identifier. - In
step 1807, the interconnect directs each of the two data containers to the appropriate adapters based on the routing information. - In step 1809(a), at least a portion of the second data container is received by
adapter 310 d, which then determines the location (e.g., ring) in the platform-memory into which the packet should be inserted. In step 1810(a),adapter 310 d performs all the necessary memory writes to insert the data payload into the appropriate memory location ofstorage 110 d. - In step 1809(b), at least a portion of the first data container is received by
adapter 310 b(1). - In step 1810(b), the
adapter 310 b(1) causes the data (i.e., the first X bytes of the protocol packet, the field-vector and unique identifier) to be stored inhost memory # 1. - Using at least some of the data received from
adapter 310 b(1), an application running on the host determines that the protocol packet should be routed to thecommunication agent 110 c(1) with a different destination address. The application does not necessarily know it holds only a part of the packet. Accordingly, the application sends the X bytes packet as if it was the entire packet. Instep 1811, thehost 202 passes the X bytes of the packet to theadapter 310 b(1) along with the unique identifier. - In
step 1812, theadapter 310 b(1) identifies that this packet is a “split” packet based on information received from the host and creates a directive. The directive includes: (1) a first MOVE instruction 1901 (seeFIG. 19 ) thatidentifiers adapter 310 c(1) as the destination of the directive, (2) amerge instruction 1902 and (3) asecond MOVE instruction 1903 thatidentifiers adapter 310 d as the destination of the directive. The adapter then passes the directive to the interconnect. - In
step 1813, the interconnect removes and executes the second (i.e., top)MOVE command 1903, thereby sending the directive toadapter 310 d. - In
step 1814,adapter 310 d receives the directive from the interconnect, removes from the received directive themerge instruction 1902 and looks up the unique identifier (e.g., look the identifier up in a memory ring mapping table). Theadapter 310 d then obtains the packet from thememory agent 110 d using the unique identifier. - In
step 1815,adapter 310 d updates the header of the packet according to the merge instruction and attaches to thepacket move instruction 1901. The data container is then provided to theinterconnect 330. - In
step 1816, theinterconnect 330 pops the MOVE instruction contained in the directive of the data container and directs the data payload of the data container (i.e., the protocol packet) to adapter identified in the MOVE instruction (which, in this case isadapter 310 c(1)). - In
step 1817, theadapter 310 c(1) provides the protocol packet to thecommunication agent 110 c(1), which may then transmit the packet over a communication channel. - As illustrated by the above example, the capabilities of
PH 320 can improve performance ofdevice 100 by preventing unnecessary data from getting to the host memory. As illustrated above, the host can perform a routing decision for a packet without needing the entire packet. For example, the host needs only a finite set of fields upon which it routes the packet. Since the uplink to the host (and from the host to it's memory) is a bottleneck in the system, saving data transfers on this bus can improve the entire platform performance. - Referring now to
FIGS. 20A ,B and 21A,B,FIGS. 20A ,B show an example implementation of the data flow shown inFIG. 17 andFIGS. 21A ,B is a flow chart describing the steps of the example implementation. - Referring now to
FIG. 21A , the data flow may begin instep 2101, whereincommunication agent 110 c(1) receives a protocol packet to be delivered toprocessing agent 110 b(1). Instep 2102, the interface (e.g. Ethernet Mac) translates the information from 1st and/or 2nd layers. Instep 2103, the packet is passed to the DMA module, which in a simple case scenario doesn't need to perform any batch operation and simply passes the packet to the classifier. - In
step 2104, classifier parses the protocol packet and determines the stream to which the packet belongs (e.g., it may identify that the packet as belonging to a certain stream (e.g., stream #N)) and determines, based on the determined stream, whether the protocol packet should be “split.” A split operation means that, for example, certain data and/or only a portion (e.g., the first X number of bytes of the packet, where X is greater than or equal to 0, along with fields that were gathered by the classifier/parser and inserted as directives of extra-info to the processor) need be provided toprocessing agent 110 b(1), while a copy of the entire packet or the portion of the packet that was not sent to the processing agent should be copied tostorage agent 110 d (which may be referred to as “platform memory”). For this example, we shall assume the packet should be split. - In
step 2105, the classifier builds a vector of fields. This field-vector may include fields that were extracted from the packet header (e.g., source/dest addresses eth/ip etc.), fields extracted from the packet data, and/or data resulting from a certain operation (e.g. tupliz-hash calculation on the packet data/header-fields) or retrieved from storage (e.g., from stored configuration information). - In
step 2106, the classifier sends a “split” command to the action engine along with the entire protocol packet and the field-vector. - In
step 2107, the action engine creates a first data container that includes a data payload and a directive. The data payload includes the first X bytes of the protocol packet. The directive includes: (1) routing information that instructsinterconnect 330 to provide the data container toadapter 310 b(1), (2) the field-vector and (3) a unique identifier that represents the protocol packet and is used, in a later phase, to retrieve the protocol packet. - In
step 2108, the action engine creates a second data container that includes a data payload and a directive. The data payload includes the entire application data portion of the protocol packet (it may also include the headers). The directive includes: (1) routing information that instructs interconnect to provide the data container toadapter 310 d and (2) the unique identifier. - In
step 2109, the interconnect directs each of the two data containers to theappropriate adapters 310 based on the routing information. - In step 2110(a), at least a portion of the second data container is received by the TX classifier of
adapter 310 d, which then determines the location (e.g., ring) in the platform-memory into which the packet should be inserted and provides the location information and the data payload to the action engine. - In step 2111(a), the action engine passes the data payload to the DMA along with the location information determined by the classifier.
- In step 2112(a), the DMA receives the data payload and the location information, and performs all the necessary memory writes through the interface (which may resemble a Memory Controller) to insert the data payload into the appropriate memory location.
- In step 2110(b), at least a portion of the first data container is received by the TX classifier of
adapter 310 b and the classifier passes a command to the action engine to send the data received by the classifier to Hostmemory # 1. - In step 2111(b), the action engine removes from the directive the field-vector and passes it (e.g., out-of-band) and the data payload to the DMA.
- In step 2112(b), the DMA provides to the agent interface a request to store the data payload the field-vector, and unique identifier in
host memory # 1. The field-vector is sent to a predefined descriptor area inHost memory # 1. - In step 2113 (see
FIG. 21B ), the agent interface translates the request to the host specific bus. Instep 2114, the data (i.e., the first X bytes of the protocol packet, the field-vector and unique identifier) is stored inhost memory # 1. - Using the field-vector it received and the first X bytes from the protocol packet, an application running on the host determines that the protocol packet should be routed to the
communication agent 110 c(1) with a different destination address. The application does not necessarily know it holds only a part of the packet. Accordingly, it sends the X bytes packet as if it was the entire packet. - In
step 2115, the host passes the X bytes of the packet to the DMA through the agent interface along with a descriptor that includes the unique identifier. - In step 2116, the agent interface passes the packet and it's descriptor to the DMA which passes the information to the classifier.
- In
step 2117, the classifier identifies that this packet is a “split” packet based on information received from the host and then passes a “merge” command to the action engine. The merge command includes: anidentifier identifying adapter 310 d, anidentifier identifying adapter 310 c(1), and the unique identifier. - In
step 2118, the action engine creates a directive. The directive includes: (1) a first MOVE instruction that identifiersadapter 310 c(1) as the destination of the directive, (2) a merge instruction and (3) a second MOVE instruction that identifiersadapter 310 d as the destination of the directive. The action engine then passes the directive to the interconnect. - In
step 2119, the interconnect removes and executes the second (i.e., top) MOVE command, thereby sending the directive toadaptor 310 d. - In
step 2120, the TX classifier ofadapter 310 d receives the directive from the interconnect, removes from the received directive the merge instruction and looks up the unique identifier (e.g., look the identifier up in a memory ring mapping table). The classifier then instructs the DMA to obtain the packet from thememory agent 110 d. - In
step 2121, the TX classifier also provides a merge request to the RX classifier with the unique identifier that was extracted from the merge instruction and provides the directive to the RX classifier, which now only includes the first MOVE instruction. The RX classifier awaits the full packet to return from the DMA. - In
step 2122, the DMA issues a set of read operations in order to receive the packet from thememory agent 110 d. - In
step 2123, the DMA passes the full packet along with its unique identifier to the RX classifier. - In
step 2124, the RX classifier passes to the action engine: (1) the directive it received from the TX classifier (2) a packet update command, and (3) the full packet. The packet update command includes the fields (e.g., IP and/or Ethernet destination address) that should be replaced and their new values. - In
step 2125, the action engine updates the header of the full packet according to the update command it received from the classifier and attaches to the packet the directive it received, thereby creating a data container. The data container is then provided to the interconnect. - In
step 2126, the interconnect pops the MOVE instruction contained in the directive of the data container and directs the data payload of the data container (i.e., the protocol packet) to adapter identified in the MOVE instruction (which, in this case isadapter 310 c(1)). - In
step 2127, theadapter 310 c(1) provides the protocol packet to thecommunication agent 110 c(1), which may then transmit the packet over a communication channel. - Referring now to
FIG. 23 ,FIG. 23 illustrates asystem 2300 according to some embodiments of the invention. As illustrated inFIG. 23 , the above described features of the present invention can be used to enable the creation of a scalablemulti-node interconnect network 2300. Insystem 2300, aplatform hub 320 and its associated agents are considered to be a “node.” In the example shown inFIG. 23 ,system 2300 includes two nodes (i.e.,node # 1 and node #2) that interconnect with each other usingscalability adapter units 2302. - While
FIG. 23 illustrates the scalability adapter unit ofnode # 1 being connected to the scalability adapter unit ofnode # 2, this was done merely for illustration. It is contemplated that an interconnect may be positioned between the scalability adapter units, thereby enabling a scalability adapter unit to communicate with several other scalability adapter units. - There can be several ways to implement a multi-node network, such as
network 2300. An example of a simple implementation would be to add a node identifier to a MOVE instruction. In addition, in this kind of implementation, DMA translation tables may have to be enlarged to include its visible ports in each node. Eachinterconnect 330 may hold a structure that indicates thecorrect scalability CAU 2302 to which to send each packet in order to get to the required node. Each interconnect routing structure is configured to forward each packet to the correct node. A packet that reaches its destination node is routed by the interconnect of the destination node to the correct port on that node. In some embodiments, aninterconnect 330 pops the head MOVE command only if the node identifier included in the MOVE command identifies the node of the interconnect. - The
scalability CAUs 2302 enable the connection between the different interconnects, thereby connecting different nodes. The scalability CAU can be configured to forward multicast messages to the next node or restrict them to the nodes boundary. - Each communication CAU (i.e., a CAU that serves a communication agent) can be further configured to enable a fail-proof structure. In case of a fail or high-load situation on the nodes processing agents, a communication CAU may forward new incoming streams to a different node. This feature is illustrated in
FIG. 24 . As illustrated inFIG. 24 ,CAU 310 c(1) ofnode # 2 may receive a data fromagent 110 c(2) ofnode # 2 and, instead of sending the data to a processing agent ofnode # 2, may send the data to a processing agent onnode # 1 by, for example, creating a data container that contains the data and a directive that includes a MOVE command that includes an identifier identifyingnode # 1, which move command causes theinterconnect 330 ofnode # 2 to send the data container to thescalability unit 2302 ofnode # 2. Upon receiving the data container, thescalability unit 2302 ofnode # 2 sends the data container to an adapter unit ofnode # 1. This adapter unit ofnode # 1 may send the data to a processing agent onnode # 1, which processing agent may process the data and then send the processed data to a communication agent onnode # 1. In this manner, a communication stream from one node is directed to a processing agent in another node and then forwarded to a different communication agent. - Using several platform hubs in a scalable connection enables multiple connections to several processors and gives the ability to spread traffic from a single communication agent to a scalable number of processing agents. This feature is illustrated in
FIG. 25 . As illustrated inFIG. 25 ,CAU 310 c(1) may be configured so that some data received fromagent 110 c(1) is provided to a processing agent ofnode # 2, whereas some other data received fromagent 110 c(1) is provided to a processing agent ofnode # 1. For example,CAU 310 c(1) may be configured so that protocol packets received fromagent 110 c(1) having a certain characteristic (e.g., source address or other characteristic) are provided to a processing agent onnode # 1, whereas other protocol packets are provided to a processing agent ofnode # 2. In this manner, traffic can be spread to a scalable number of processing agents. - Additionally, using several platform hubs in a scalable connection enables multiple connections to a communication agent. This means the communication channel bandwidth is scalable as well. This feature is illustrated in
FIG. 26 . - While various embodiments/variations of the present invention have been described above, it should be understood that they have been presented by way of example only, and not limitation. Thus, the breadth and scope of the present invention should not be limited by any of the above-described exemplary embodiments.
- Additionally, while the processes described above and illustrated in the drawings are shown as a sequence of steps, this was done solely for the sake of illustration. Accordingly, it is contemplated that some steps may be added, some steps may be omitted, and the order of the steps may be re-arranged.
Claims (20)
1. A processing device, comprising:
a housing;
a chip housed in the housing;
a first processing agent housed in the housing and connected to the chip; and
a network interface operable to connect the processing device to an external network that is external to the housing, wherein
the chip comprises:
an interconnect;
a second processing agent communicatively connected to the interconnect, the second processing agent comprising a plurality of general purpose processor cores and a cache memory; and
a network adapter communicatively connected to the interconnect and to the network interface, the network adapter being operable to receive and classify packets received from the network interface, wherein
the network adapter is configured such that, in response to receiving a particular packet, the network adapter uses the interconnect to store a payload portion of the packet in the cache memory of the second processing agent.
2. The processing device of claim 1 , wherein the network adapter is configured such that, in response to receiving via the network interface a packet having a header and a data payload, the network adapter i) compares data from both the data payload and header with pre-determined data, and ii) performs a certain predetermined action in response to determining that the compared data matches the predetermined data.
3. The processing device of claim 2 , wherein the network adapter comprises a packet parser and an action engine communicatively connected to the packet parser, wherein the action engine is operable to receive from the packet parser packet data and/or commands, and the action engine is further operable to do one or more of (i) perform a packet filtering operation on packets parsed by the packet parser and (ii) perform a packet modification operation on packets parsed by the packet parser.
4. The processing device of claim 1 , wherein the chip further comprises an accelerating agent communicatively connected to the interconnect, the accelerating agent being adapted to perform acceleration operations.
5. The processing device of claim 4 , wherein
the accelerating agent is a cryptographic accelerating agent, and
the second processing agent is adapted to use the interconnect to provide to the cryptographic accelerating agent i) the payload and ii) a command identifying a cryptographic operation that the cryptographic accelerating agent should perform on the payload.
6. The processing device of claim 5 , wherein the cryptographic accelerating agent comprises a direct memory access (DMA) engine coupled to the interconnect, and the DMA engine is configured to i) receive the payload provided by the second processing agent and ii) write the payload into a memory space of the cryptographic agent.
7. A chip for use in a network server comprising a first processing agent and a network interface, said chip comprising:
an interconnect;
a second processing agent communicatively connected to the interconnect, the second processing agent comprising a plurality of general purpose processor cores and a cache memory;
a network adapter communicatively connected to the interconnect and to the network interface, the network adapter being operable to receive and classify packets received from the network interface; and
an accelerating agent communicatively connected to the interconnect, the accelerating agent being adapted to perform acceleration operations.
8. The chip of claim 7 , wherein the network adapter is configured such that, in response to receiving via the network interface a packet having a header and a data payload, the network adapter i) compares data from both the data payload and header with pre-determined data, and ii) performs a certain predetermined action in response to determining that the compared data matches the predetermined data.
9. The chip of claim 8 , wherein the network adapter comprises a packet parser and an action engine communicatively connected to the packet parser, wherein the action engine is operable to receive from the packet parser packet data and/or commands, and the action engine is further operable to do one or more of (i) perform a packet filtering operation on packets parsed by the packet parser and (ii) perform a packet modification operation on packets parsed by the packet parser.
10. The chip of claim 7 , wherein the network adapter comprises a packet parser, an action engine, and an identifier module operable to receive packet information from the packet parser and, based on the packet information and configuration data, provide a command to the action engine.
11. The chip of claim 7 , wherein the network adapter is operable to determine a flow to which a packet received from the network interface.
12. The chip of claim 11 , wherein the network adapter is operable to determine whether a received packet meets specified criteria and is operable to drop the received packet in response to determining that the packet meets the specified criteria.
13. The chip of claim 11 , wherein acceleration agent is a cryptographic acceleration agent.
14. The chip of claim 7 , wherein the network adapter is further operable, such that, in response to receiving a packet from the network interface, the network adapter is operable to determine whether the received packet comprise a Transmission Control Protocol (TCP) protocol data unit (PDU).
15. A chip for use in a processing device, said chip comprising:
an interconnect;
a first processor communicatively connected to the interconnect and communicatively connected to a network interface, the first processor being operable to receive packets via the network interface; and
a second processor communicatively connected to the interconnect, the second processor comprising a plurality of processor cores and cache memory, wherein
the first processor is configured to:
parse a packet received via the network interface and perform one or more actions based on information contained in the received packet, wherein said actions include providing at least a portion of the received packet to the cache memory of the second processor via the interconnect.
16. The chip of claim 15 , wherein the cache memory comprises a plurality of L2 caches.
17. The chip of claim 15 , wherein the first processor comprises: a packet parser configured to parse packets provided to the first processor from a communication agent and to extract one or more fields from the packets; and a configurable action engine operable to (a) receive one or more packet fields extracted by the packet parser and (b) perform one or more actions based on information contained in the received one or more packet fields, wherein said actions include providing a received packet to the second processor via the interconnect.
18. The chip of claim 17 , wherein the action engine is operable to perform packet filtering of the packets received from the communication agent.
19. The chip of claim 15 , further comprising: an accelerating agent configured to perform a cryptographic function, the accelerating agent being communicatively connected to the second processor through the interconnect.
20. The chip of claim 15 , wherein the first processor is operable to determine whether a packet received via the network interface meets specified criteria and is operable to drop the packet in response to determining that the packet meets the specified criteria.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/044,688 US20160164807A1 (en) | 2007-07-11 | 2016-02-16 | Systems and Methods For Efficient Handling of Data Traffic and Processing Within a Processing Device |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/776,285 US7793032B2 (en) | 2007-07-11 | 2007-07-11 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US12/875,343 US8209457B2 (en) | 2007-07-11 | 2010-09-03 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US13/474,932 US9268729B2 (en) | 2007-07-11 | 2012-05-18 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US15/044,688 US20160164807A1 (en) | 2007-07-11 | 2016-02-16 | Systems and Methods For Efficient Handling of Data Traffic and Processing Within a Processing Device |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US13/474,932 Continuation US9268729B2 (en) | 2007-07-11 | 2012-05-18 | Systems and methods for efficient handling of data traffic and processing within a processing device |
Publications (1)
Publication Number | Publication Date |
---|---|
US20160164807A1 true US20160164807A1 (en) | 2016-06-09 |
Family
ID=40254071
Family Applications (4)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/776,285 Expired - Fee Related US7793032B2 (en) | 2007-07-11 | 2007-07-11 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US12/875,343 Expired - Fee Related US8209457B2 (en) | 2007-07-11 | 2010-09-03 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US13/474,932 Expired - Fee Related US9268729B2 (en) | 2007-07-11 | 2012-05-18 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US15/044,688 Abandoned US20160164807A1 (en) | 2007-07-11 | 2016-02-16 | Systems and Methods For Efficient Handling of Data Traffic and Processing Within a Processing Device |
Family Applications Before (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/776,285 Expired - Fee Related US7793032B2 (en) | 2007-07-11 | 2007-07-11 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US12/875,343 Expired - Fee Related US8209457B2 (en) | 2007-07-11 | 2010-09-03 | Systems and methods for efficient handling of data traffic and processing within a processing device |
US13/474,932 Expired - Fee Related US9268729B2 (en) | 2007-07-11 | 2012-05-18 | Systems and methods for efficient handling of data traffic and processing within a processing device |
Country Status (1)
Country | Link |
---|---|
US (4) | US7793032B2 (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018004896A1 (en) * | 2016-06-30 | 2018-01-04 | Intel Corporation | Graphics processing unit (gpu) for packet transfer |
Families Citing this family (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7793032B2 (en) * | 2007-07-11 | 2010-09-07 | Commex Technologies, Ltd. | Systems and methods for efficient handling of data traffic and processing within a processing device |
US8122177B1 (en) * | 2009-05-19 | 2012-02-21 | Xilinx, Inc. | Direct memory access technique for use with PCIe endpoints |
US8892712B2 (en) * | 2011-09-30 | 2014-11-18 | Oracle International Corporation | Upgrading enterprise managers |
US9237095B2 (en) * | 2012-02-17 | 2016-01-12 | Netronome Systems, Inc. | Island-based network flow processor integrated circuit |
US8902902B2 (en) | 2012-07-18 | 2014-12-02 | Netronome Systems, Incorporated | Recursive lookup with a hardware trie structure that has no sequential logic elements |
US9559897B2 (en) | 2012-12-21 | 2017-01-31 | Brocade Communications Systems, Inc. | Device ID assignment in a system of devices |
US9148387B2 (en) | 2013-05-10 | 2015-09-29 | Brocade Communications Systems, Inc. | Hardware hash table virtualization in multi-packet processor networking systems |
US9313102B2 (en) | 2013-05-20 | 2016-04-12 | Brocade Communications Systems, Inc. | Configuration validation in a mixed node topology |
US9853889B2 (en) | 2013-05-20 | 2017-12-26 | Brocade Communications Systems, Inc. | Broadcast and multicast traffic reduction in stacking systems |
US10284499B2 (en) | 2013-08-22 | 2019-05-07 | Arris Enterprises Llc | Dedicated control path architecture for systems of devices |
US9185049B2 (en) | 2013-10-31 | 2015-11-10 | Brocade Communications Systems, Inc. | Techniques for simplifying stacking trunk creation and management |
US9577932B2 (en) | 2014-02-12 | 2017-02-21 | Brocade Communications Systems, Inc. | Techniques for managing ternary content-addressable memory (TCAM) resources in heterogeneous systems |
US9692695B2 (en) | 2014-03-27 | 2017-06-27 | Brocade Communications Systems, Inc. | Techniques for aggregating hardware routing resources in a multi-packet processor networking system |
US9692652B2 (en) | 2014-04-03 | 2017-06-27 | Brocade Communications Systems, Inc. | Framework for reliably communicating port information in a system of devices |
US10091059B2 (en) | 2014-12-16 | 2018-10-02 | Arris Enterprises Llc | Handling connections between network devices that support multiple port communication modes |
EP3248360B1 (en) | 2015-01-19 | 2020-05-06 | Inauth, Inc. | Systems and methods for trusted path secure communication |
US10983920B2 (en) * | 2018-02-08 | 2021-04-20 | Xilinx, Inc. | Customizable multi queue DMA interface |
CN110492984B (en) * | 2019-07-29 | 2020-06-23 | 南方电网数字电网研究院有限公司 | Data transmission method and device, data analysis processing system and computer equipment |
US20200151362A1 (en) * | 2019-08-21 | 2020-05-14 | Intel Corporation | Integrity and data encryption (ide) over computer buses |
CN113868171A (en) * | 2021-09-28 | 2021-12-31 | 上海兆芯集成电路有限公司 | Interconnection system |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020017425A1 (en) * | 2000-07-13 | 2002-02-14 | Sang-Heon Yoon | Suction muffler of reciprocating compressor |
US20020078028A1 (en) * | 2000-12-18 | 2002-06-20 | Trevalon Inc. | Network server |
US20020174255A1 (en) * | 2001-05-18 | 2002-11-21 | Broadcom Corporation | System on a chip for packet processing |
US20030202536A1 (en) * | 2001-04-27 | 2003-10-30 | Foster Michael S. | Integrated analysis of incoming data transmissions |
US20060000786A1 (en) * | 2002-05-29 | 2006-01-05 | Ruediger Tueshaus | Filtering screen construction and methods |
US20060007860A1 (en) * | 2001-06-11 | 2006-01-12 | Bluefire Security Technologies, Inc. | Packet filtering methods and systems |
US20060020381A1 (en) * | 2004-07-24 | 2006-01-26 | Ford Motor Company | A voice notification system for a vehicle |
US20060026378A1 (en) * | 2004-07-27 | 2006-02-02 | Somsubhra Sikdar | Array machine context data memory |
US20060075067A1 (en) * | 2004-08-30 | 2006-04-06 | International Business Machines Corporation | Remote direct memory access with striping over an unreliable datagram transport |
US20080002682A1 (en) * | 2006-06-30 | 2008-01-03 | Sun Microsystems, Inc. | Generalized serialization queue framework for protocol processing |
US20080240140A1 (en) * | 2007-03-29 | 2008-10-02 | Microsoft Corporation | Network interface with receive classification |
US9268729B2 (en) * | 2007-07-11 | 2016-02-23 | Commex Technologies, Ltd. | Systems and methods for efficient handling of data traffic and processing within a processing device |
Family Cites Families (55)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2567922B2 (en) * | 1988-08-30 | 1996-12-25 | 株式会社日立製作所 | Path control method |
US5206939A (en) * | 1990-09-24 | 1993-04-27 | Emc Corporation | System and method for disk mapping and data retrieval |
JP3290302B2 (en) | 1994-06-30 | 2002-06-10 | 富士通株式会社 | Data exchange processing method and apparatus |
US5561669A (en) | 1994-10-26 | 1996-10-01 | Cisco Systems, Inc. | Computer network switching system with expandable number of ports |
US5761516A (en) * | 1996-05-03 | 1998-06-02 | Lsi Logic Corporation | Single chip multiprocessor architecture with internal task switching synchronization bus |
JP3343054B2 (en) * | 1997-07-01 | 2002-11-11 | ケイディーディーアイ株式会社 | Internet compatible link monitor method |
SE520101C2 (en) * | 1998-05-13 | 2003-05-27 | Axis Ab | Integrated circuit and method to induce an integrated circuit to execute instructions |
US6904519B2 (en) * | 1998-06-12 | 2005-06-07 | Microsoft Corporation | Method and computer program product for offloading processing tasks from software to hardware |
US7346063B1 (en) | 1998-07-08 | 2008-03-18 | Broadcom Corporation | Memory management unit for a network switch |
US6370583B1 (en) * | 1998-08-17 | 2002-04-09 | Compaq Information Technologies Group, L.P. | Method and apparatus for portraying a cluster of computer systems as having a single internet protocol image |
US6359886B1 (en) * | 1998-08-17 | 2002-03-19 | Compaq Computer Corporation | Method and apparatus for filtering and routing communications frames |
US6542961B1 (en) * | 1998-12-22 | 2003-04-01 | Hitachi, Ltd. | Disk storage system including a switch |
US6788704B1 (en) * | 1999-08-05 | 2004-09-07 | Intel Corporation | Network adapter with TCP windowing support |
US6768742B1 (en) | 1999-10-08 | 2004-07-27 | Advanced Micro Devices, Inc. | On-chip local area network |
US7031258B1 (en) | 2000-01-13 | 2006-04-18 | Mercury Computer Systems, Inc. | Digital data system with link level message flow control |
US6993621B1 (en) * | 2000-03-31 | 2006-01-31 | Emc Corporation | Data storage system having separate data transfer section and message network with plural directors on a common printed circuit board and redundant switching networks |
US7032031B2 (en) | 2000-06-23 | 2006-04-18 | Cloudshield Technologies, Inc. | Edge adapter apparatus and method |
US7114008B2 (en) | 2000-06-23 | 2006-09-26 | Cloudshield Technologies, Inc. | Edge adapter architecture apparatus and method |
US6775719B1 (en) | 2000-09-28 | 2004-08-10 | Intel Corporation | Host-fabric adapter and method of connecting a host system to a channel-based switched fabric in a data network |
US20020071450A1 (en) | 2000-12-08 | 2002-06-13 | Gasbarro Dominic J. | Host-fabric adapter having bandwidth-optimizing, area-minimal, vertical sliced memory architecture and method of connecting a host system to a channel-based switched fabric in a data network |
US20020199014A1 (en) * | 2001-03-26 | 2002-12-26 | Accton Technology Corporation | Configurable and high-speed content-aware routing method |
US6934702B2 (en) * | 2001-05-04 | 2005-08-23 | Sun Microsystems, Inc. | Method and system of routing messages in a distributed search network |
US6904057B2 (en) * | 2001-05-04 | 2005-06-07 | Slt Logic Llc | Method and apparatus for providing multi-protocol, multi-stage, real-time frame classification |
US7082502B2 (en) | 2001-05-15 | 2006-07-25 | Cloudshield Technologies, Inc. | Apparatus and method for interfacing with a high speed bi-directional network using a shared memory to store packet data |
US7260616B1 (en) | 2001-08-13 | 2007-08-21 | Sprint Communications Company L.P. | Communication hub with automatic device registration |
US7133365B2 (en) | 2001-11-02 | 2006-11-07 | Internap Network Services Corporation | System and method to provide routing control of information over networks |
US7640347B1 (en) * | 2002-05-02 | 2009-12-29 | F5 Networks, Inc. | Method and system for inserting POST data into the GET request to apply normal caching rules |
US7224670B2 (en) * | 2002-06-03 | 2007-05-29 | International Business Machines Corporation | Flow control in computer networks |
US7051150B2 (en) | 2002-07-29 | 2006-05-23 | Freescale Semiconductor, Inc. | Scalable on chip network |
US7277449B2 (en) | 2002-07-29 | 2007-10-02 | Freescale Semiconductor, Inc. | On chip network |
US7254114B1 (en) | 2002-08-26 | 2007-08-07 | Juniper Networks, Inc. | Network router having integrated flow accounting and packet interception |
US7251215B1 (en) | 2002-08-26 | 2007-07-31 | Juniper Networks, Inc. | Adaptive network router |
US20040153460A1 (en) * | 2003-01-30 | 2004-08-05 | International Business Machines Corporation | Reduction of ternary rules with common priority and actions |
US7274666B2 (en) * | 2003-04-01 | 2007-09-25 | International Business Machines Corporation | Method and system for managing traffic within a data communication network |
US7293113B1 (en) | 2003-05-28 | 2007-11-06 | Advanced Micro Devices, Inc. | Data communication system with hardware protocol parser and method therefor |
US7483374B2 (en) * | 2003-08-05 | 2009-01-27 | Scalent Systems, Inc. | Method and apparatus for achieving dynamic capacity and high availability in multi-stage data networks using adaptive flow-based routing |
US7447240B2 (en) | 2004-03-29 | 2008-11-04 | Micron Technology, Inc. | Method and system for synchronizing communications links in a hub-based memory system |
US7210000B2 (en) | 2004-04-27 | 2007-04-24 | Intel Corporation | Transmitting peer-to-peer transactions through a coherent interface |
US7860096B2 (en) | 2004-06-08 | 2010-12-28 | Oracle America, Inc. | Switching method and apparatus for use in a communications network |
US7284082B2 (en) | 2004-08-19 | 2007-10-16 | Lsi Corporation | Controller apparatus and method for improved data transfer |
US7249207B2 (en) * | 2005-01-31 | 2007-07-24 | International Business Machines Corporation | Internal data bus interconnection mechanism utilizing central interconnection module converting data in different alignment domains |
GB2424144B (en) * | 2005-03-11 | 2008-12-31 | 3Com Corp | Packet diversion in switching fabrics and multiple forwarding instructions for packets |
JP4316522B2 (en) | 2005-03-15 | 2009-08-19 | サンデン株式会社 | Connection device for communication equipment |
US7554928B2 (en) * | 2005-04-01 | 2009-06-30 | Cisco Technology, Inc. | Clustering methods for scalable and bandwidth-efficient multicast |
US7617436B2 (en) * | 2005-08-02 | 2009-11-10 | Nokia Corporation | Method, device, and system for forward channel error recovery in video sequence transmission over packet-based network |
ATE379918T1 (en) * | 2005-11-18 | 2007-12-15 | Alcatel Lucent | METHOD FOR PROVIDING SERVICES FROM A SERVER TO A CLIENT OVER A MOBILE WIRELESS COMMUNICATIONS NETWORK, CORRESPONDING NODE AND TERMINAL |
US7430633B2 (en) * | 2005-12-09 | 2008-09-30 | Microsoft Corporation | Pre-storage of data to pre-cached system memory |
JP2007286806A (en) * | 2006-04-14 | 2007-11-01 | Hitachi Ltd | Storage system and data storage method |
US20070291782A1 (en) * | 2006-06-14 | 2007-12-20 | Claude Basso | Acknowledgement filtering |
US8050266B2 (en) * | 2006-07-20 | 2011-11-01 | Oracle America, Inc. | Low impact network debugging |
US7487284B2 (en) | 2006-07-28 | 2009-02-03 | Intel Corporation | Transaction flow and ordering for a packet processing engine, located within an input-output hub |
US8041854B2 (en) * | 2007-09-28 | 2011-10-18 | Intel Corporation | Steering data units to a consumer |
US8509237B2 (en) * | 2009-06-26 | 2013-08-13 | Wisconsin Alumni Research Foundation | Architecture and system for coordinated network-wide redundancy elimination |
US20110228674A1 (en) * | 2010-03-18 | 2011-09-22 | Alon Pais | Packet processing optimization |
US9450780B2 (en) * | 2012-07-27 | 2016-09-20 | Intel Corporation | Packet processing approach to improve performance and energy efficiency for software routers |
-
2007
- 2007-07-11 US US11/776,285 patent/US7793032B2/en not_active Expired - Fee Related
-
2010
- 2010-09-03 US US12/875,343 patent/US8209457B2/en not_active Expired - Fee Related
-
2012
- 2012-05-18 US US13/474,932 patent/US9268729B2/en not_active Expired - Fee Related
-
2016
- 2016-02-16 US US15/044,688 patent/US20160164807A1/en not_active Abandoned
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020017425A1 (en) * | 2000-07-13 | 2002-02-14 | Sang-Heon Yoon | Suction muffler of reciprocating compressor |
US20020078028A1 (en) * | 2000-12-18 | 2002-06-20 | Trevalon Inc. | Network server |
US20030202536A1 (en) * | 2001-04-27 | 2003-10-30 | Foster Michael S. | Integrated analysis of incoming data transmissions |
US20020174255A1 (en) * | 2001-05-18 | 2002-11-21 | Broadcom Corporation | System on a chip for packet processing |
US20060007860A1 (en) * | 2001-06-11 | 2006-01-12 | Bluefire Security Technologies, Inc. | Packet filtering methods and systems |
US20060000786A1 (en) * | 2002-05-29 | 2006-01-05 | Ruediger Tueshaus | Filtering screen construction and methods |
US20060020381A1 (en) * | 2004-07-24 | 2006-01-26 | Ford Motor Company | A voice notification system for a vehicle |
US20060026378A1 (en) * | 2004-07-27 | 2006-02-02 | Somsubhra Sikdar | Array machine context data memory |
US20060075067A1 (en) * | 2004-08-30 | 2006-04-06 | International Business Machines Corporation | Remote direct memory access with striping over an unreliable datagram transport |
US20080002682A1 (en) * | 2006-06-30 | 2008-01-03 | Sun Microsystems, Inc. | Generalized serialization queue framework for protocol processing |
US20080240140A1 (en) * | 2007-03-29 | 2008-10-02 | Microsoft Corporation | Network interface with receive classification |
US9268729B2 (en) * | 2007-07-11 | 2016-02-23 | Commex Technologies, Ltd. | Systems and methods for efficient handling of data traffic and processing within a processing device |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018004896A1 (en) * | 2016-06-30 | 2018-01-04 | Intel Corporation | Graphics processing unit (gpu) for packet transfer |
US10331590B2 (en) | 2016-06-30 | 2019-06-25 | Intel Corporation | Graphics processing unit (GPU) as a programmable packet transfer mechanism |
Also Published As
Publication number | Publication date |
---|---|
US9268729B2 (en) | 2016-02-23 |
US7793032B2 (en) | 2010-09-07 |
US8209457B2 (en) | 2012-06-26 |
US20090019206A1 (en) | 2009-01-15 |
US20130016729A1 (en) | 2013-01-17 |
US20100332713A1 (en) | 2010-12-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9268729B2 (en) | Systems and methods for efficient handling of data traffic and processing within a processing device | |
US10997106B1 (en) | Inter-smartNIC virtual-link for control and datapath connectivity | |
US10742779B2 (en) | Configurable parser and a method for parsing information units | |
JP3645734B2 (en) | Network relay device and network relay method | |
US6683885B1 (en) | Network relaying apparatus and network relaying method | |
US7742474B2 (en) | Virtual network interface cards with VLAN functionality | |
US7630368B2 (en) | Virtual network interface card loopback fastpath | |
US7474666B2 (en) | Switch port analyzers | |
US8904028B2 (en) | Scalable cluster router | |
US8194667B2 (en) | Method and system for inheritance of network interface card capabilities | |
US8949578B2 (en) | Sharing of internal pipeline resources of a network processor with external devices | |
US9300597B2 (en) | Statistics module for network processors in virtual local area networks | |
US20080019359A1 (en) | Multiple virtual network stack instances using virtual network interface cards | |
US20140016648A1 (en) | Communication control system, switch node and communication control method | |
US7613132B2 (en) | Method and system for controlling virtual machine bandwidth | |
TW201519607A (en) | Semiconductor with virtualized computation and switch resources | |
US11394664B2 (en) | Network interface device | |
US9590922B2 (en) | Programmable and high performance switch for data center networks | |
CN1988528A (en) | Dynamic services blade and method | |
US20220206957A1 (en) | Methods and systems for using a packet processing pipeline to accelerate infiniband administrative operations | |
US20080043632A1 (en) | Low impact network debugging | |
JP3645735B2 (en) | Network relay device and network relay method | |
JP3645733B2 (en) | Network relay device and network relay method | |
EP3491792B1 (en) | Deliver an ingress packet to a queue at a gateway device | |
US11757778B2 (en) | Methods and systems for fairness across RDMA requesters using a shared receive queue |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STPP | Information on status: patent application and granting procedure in general |
Free format text: ADVISORY ACTION MAILED |
|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION |