Abstract: An originating host device in a distribution chain is provided upstream from multiple host devices including intermediary and terminating host devices. The originating host device includes a core with a generation application and a first RDMA NIC. The core: determines a plan for transferring data between the originating host device and the other host devices; and generates WQEs to implement the plan. The first RDMA NIC includes a read application, a descriptor application, and a reception application. The read application is configured such that the first RDMA NIC reads the WQEs from a first memory. The descriptor application is configured such that the first RDMA NIC: writes a portion of the WQEs directly from the first RDMA NIC to a second memory of the intermediary host device; and triggers a second RDMA NIC of the intermediary host device to process the portion of the WQEs stored in the second memory.

Abstract: Distributed storage systems, devices, and associated methods of data replication are disclosed herein. In one embodiment, a server in a distributed storage system is configured to write, with an RDMA enabled NIC, a block of data from a memory of the server to a memory at another server via an RDMA network. Upon completion of writing the block of data to the another server, the server can also send metadata representing a memory location and a data size of the written block of data in the memory of the another server via the RDMA network. The sent metadata is to be written into a memory location containing data representing a memory descriptor that is a part of a data structure representing a pre-posted work request comgured to write a copy of the block of data from the another server to an additional server via the RDMA network.

Abstract: Data processing apparatus includes a host processor and a network interface controller (NIC), which is configured to couple the host processor to a packet data network. A memory holds a flow state table containing context information with respect to computational operations to be performed on multiple packet flows conveyed between the host processor and the network. Acceleration logic is coupled to perform the computational operations on payloads of packets in the multiple packet flows using the context information in the flow state table.

Abstract: Techniques for facilitating load balancing in distributed computing systems are disclosed herein. In one embodiment, a method includes receiving, at a destination server, a request packet from a load balancer via the computer network requesting a remote direct memory access (“RDMA”) connection between an originating server and one or more other servers selectable by the load balancer. The method can also include configuring, at the destination server, a rule for processing additional packets transmittable to the originating server via the RDMA connection based on the received reply packet. The rule is configured to encapsulate an outgoing packet transmittable to the originating server with an outer header having a destination field containing a network address of the originating server and a source field containing another network address of the destination server.

Abstract: Techniques are disclosed for identifying faulty links in a virtualized computing environment. Network path latency information is received for one or more network paths in the networked computing environment. Based on the network path latency information, a probable presence of a faulty component is determined. In response to the determination, physical links for a network path associated with the probable faulty component are identified. Information indicative of likely sources of the probable faulty component is received from multiple hosts of the networked computing environment. Based on the identified physical links and information, a faulty component is determined.

Abstract: Distributed storage systems, devices, and associated methods of data replication are disclosed herein. In one embodiment, a server in a distributed storage system is configured to write, with an RDMA enabled NIC, a block of data from a memory of the server to a memory at another server via an RDMA network. Upon completion of writing the block of data to the another server, the server can also send metadata representing a memory location and a data size of the written block of data in the memory of the another server via the RDMA network. The sent metadata is to be written into a memory location containing data representing a memory descriptor that is a part of a data structure representing a pre-posted work request configured to write a copy of the block of data from the another server to an additional server via the RDMA network.

Abstract: Distributed storage systems, devices, and associated methods of data replication are disclosed herein. In one embodiment, a server in a distributed storage system is configured to write, with an RDMA enabled NIC, a block of data from a memory of the server to a memory at another server via an RDMA network. Upon completion of writing the block of data to the another server, the server can also send metadata representing a memory location and a data size of the written block of data in the memory of the another server via the RDMA network. The sent metadata is to be written into a memory location containing data representing a memory descriptor that is a part of a data structure representing a pre-posted work request configured to write a copy of the block of data from the another server to an additional server via the RDMA network.

Abstract: A method for interaction by a central processing unit (CPU) and peripheral devices in a computer includes allocating, in a memory, a work queue for controlling a first peripheral device of the computer. The CPU prepares a work request for insertion in the allocated work queue, the work request specifying an operation for execution by the first peripheral device. A second peripheral device of the computer submits an instruction to the first peripheral device to execute the work request that was prepared by the CPU and thereby to perform the operation specified by the work request.

Abstract: Computing apparatus includes a central processing unit (CPU), which is configured to run concurrently multiple virtual machines, including at least first and second virtual machines. A peripheral component bus is connected to communicate with the CPU. Multiple peripheral devices are connected to communicate via the bus with the CPU and with others of the peripheral devices, including at least first and second peripheral devices that are each respectively partitioned into at least first and second functional entities, which are respectively assigned to serve the at least first and second virtual machines. Access control logic is configured to forward peer-to-peer communications initiated by the functional entities between the peripheral devices over the bus while inhibiting access in the peer-to-peer communications between the functional entities that are assigned to different ones of the virtual machines.

Abstract: A method for computing includes submitting a first command from a central processing unit (CPU) to a first peripheral device in a computer to write data in a first bus transaction over a peripheral component bus in the computer to a second peripheral device in the computer. A second command is submitted from the CPU to one of the first and second peripheral devices to execute a second bus transaction, subsequent to the first bus transaction, that will flush the data from the peripheral component bus to the second peripheral device. The first and second bus transactions are executed in response to the first and second commands. Following completion of the second bus transaction, the second peripheral device processes the written data in.

Abstract: A method in a network node that includes a host and an accelerator, includes holding a work queue that stores work elements, a notifications queue that stores notifications of the work elements, and control indices for adding and removing the work elements and the notifications to and from the work queue and the notifications queue, respectively. The notifications queue resides on the accelerator, and at least some of the control indices reside on the host. Messages are exchanged between a network and the network node using the work queue, the notifications queue and the control indices.

Abstract: Peripheral apparatus for use with a host computer includes an add-on device, which includes a first network port coupled to one end of a packet communication link and add-on logic, which is configured to receive and transmit packets containing data over the packet communication link and to perform computational operations on the data. A network interface controller (NIC) includes a host bus interface, configured for connection to the host bus of the host computer and a second network port, coupled to the other end of the packet communication link. Packet processing logic in the NIC is coupled between the host bus interface and the second network port, and is configured to translate between the packets transmitted and received over the packet communication link and transactions executed on the host bus so as to provide access between the add-on device and the resources of the host computer.

Abstract: A data processing device includes a first packet communication interface for communication with at least one host processor via a network interface controller (NIC) and a second packet communication interface for communication with a packet data network. A memory holds a flow state table containing context information with respect to multiple packet flows conveyed between the host processor and the network via the first and second interfaces packet communication interfaces. Acceleration logic, coupled between the first and second packet communication interfaces, performs computational operations on payloads of packets in the multiple packet flows using the context information in the flow state table.

Abstract: A method includes communicating between at least first and second devices over a bus in accordance with a bus address space, including providing direct access over the bus to a local address space of the first device by mapping at least some of the addresses of the local address space to the bus address space. In response to indicating, by the first device or the second device, that the second device requires to access a local address in the local address space that is not currently mapped to the bus address space, the local address is mapped to the bus address space, and the local address is accessed directly, by the second device, using the mapping.

Abstract: Techniques for facilitating load balancing in distributed computing systems are disclosed herein. In one embodiment, a method includes receiving, at a destination server, a request packet from a load balancer via the computer network requesting a remote direct memory access (“RDMA”) connection between an originating server and one or more other servers selectable by the load balancer. The method can also include configuring, at the destination server, a rule for processing additional packets transmittable to the originating server via the RDMA connection based on the received reply packet. The rule is configured to encapsulate an outgoing packet transmittable to the originating server with an outer header having a destination field containing a network address of the originating server and a source field containing another network address of the destination server.

Abstract: Communication apparatus includes a plurality of interfaces configured to be connected to a Layer-3 packet network and to serve as ingress and egress interfaces to receive and transmit packets from and to the network. Routing logic is coupled to process respective Layer-3 headers of the packets received through the ingress interfaces and to route the packets via the egress interfaces to respective destinations indicated by the Layer-3 headers. Congestion detection logic is coupled to identify a flow of the received packets that is causing congestion in the network and a Layer-3 address from which the flow originates, and to direct the routing logic to route a backward congestion notification message (CNM) packet via one of the egress interfaces to the identified Layer-3 address.

Abstract: Systems and methods for automatic generation of an exception detector for determining an overflow condition are provided. In one example, a method including receiving code corresponding to at least one block of the hardware module, wherein the code providing at least a first value for a first variable, a second value for a second variable, or an operation to be performed using at least one of the first variable or the second variable, is provided. The method may further include automatically generating an exception detector to determine whether an overflow condition is triggered by at least one of the operation, the first variable, or the second variable.

Abstract: An originating host device in a distribution chain is provided upstream from multiple host devices including intermediary and terminating host devices. The originating host device includes a core with a generation application and a first RDMA NIC. The core: determines a plan for transferring data between the originating host device and the other host devices; and generates WQEs to implement the plan. The first RDMA NIC includes a read application, a descriptor application, and a reception application. The read application is configured such that the first RDMA NIC reads the WQEs from a first memory. The descriptor application is configured such that the first RDMA NIC: writes a portion of the WQEs directly from the first RDMA NIC to a second memory of the intermediary host device; and triggers a second RDMA NIC of the intermediary host device to process the portion of the WQEs stored in the second memory.

Abstract: A method in a network element that includes multiple interfaces for connecting to a communication network includes receiving from the communication network via an ingress interface a flow including a sequence of packets, and routing the packets to a destination of the flow via a first egress interface. A permission indication for re-routing the flow is received in the ingress interface. In response to receiving the permission indication, subsequent packets of the flow are re-routed via a second egress interface that is different from the first egress interface. Further re-routing of the flow is refrained from, until receiving another permission indication.

Abstract: A method for memory access is applied in a cluster of computers linked by a network. For a given computer, a respective physical memory range is defined including a local memory range within the local RAM of the given computer and a remote memory range allocated to the given compute within the local RAM of at least one other computer in the cluster, which is accessible via the network using the network interface controllers of the computers. When a memory operation is requested at a given address in the respective physical memory range, the operation is executed on the data in the local RAM of the given computer when the data at the given address are valid in the local memory range. Otherwise the data are fetched from the given address in the remote memory range to the local memory range before executing the operation on the data.