Intel Corporation

Netapp, Inc.

Solidfire, Inc.

Peter P Waskiewicz, Jr

Eliezer Tamir

Iosif Gasparakis

Andrey Chilikin

Andrew Cunningham

Johannes Berg

Marshall McMullen

Patrick L Connor

Parthasarathy Sarangam

Anjali Singhai Jain

Neerav Parikh

Scott P Dubal

Jared Cantwell

Derek John Leslie

Brian L Kantor

Carl Jeffrey Seelye

Aaron K Salter

Technologies for programming flexible accelerated network pipelines include a comping device with a network controller. The computing device loads a program binary file that includes a packet processing program and a requested hint section. The binary file may be an executable and linkable format (ELF) file with an extended Berkeley packet filter (eBPF) program. The computing device determines a hardware configuration for the network controller based on the requested offload hints and programs the network controller. The network controller processes network packets with the requested offloads, such as packet classification, hashing, checksums, traffic shaping, or other offloads. The network controller returns results of the offloads as hints in metadata. The packet processing program performs actions based on the metadata, such as forwarding, dropping, packet modification, or other actions. The computing device may compile an eBPF source file to generate the binary file. Other embodiments are described and claimed.

Technologies for programming flexible accelerated network pipeline using EBPF

Examples include registering a device driver with an operating system, including registering available hardware offloads. The operating system receives a call to a hardware offload, inserts a binary filter representing the hardware offload into a hardware component and causes the execution of the binary filter by the hardware component when the hardware offload is available, and executes the binary filter in software when the hardware offload is not available.

Extending Berkeley packet filter semantics for hardware offloads

Technologies for programming flexible accelerated network pipeline using eBPF

Extending Berkeley Packet Filter semantics for hardware offloads

Extending berkeley packet filter semantics for hardware offloads

Methods, apparatus, and systems for implementing in Network Interface Controller (NIC) flow switching. Switching operations are effected via hardware-based forwarding mechanisms in apparatus such as NICs in a manner that does not employ use of computer system processor resources and is transparent to operating systems hosted by such computer systems. The forwarding mechanisms are configured to move or copy Media Access Control (MAC) frame data between receive (Rx) and transmit (Tx) queues associated with different NIC ports that may be on the same NIC or separate NICs. The hardware-based switching operations effect forwarding of MAC frames between NIC ports using memory operations, thus reducing external network traffic, internal interconnect traffic, and processor workload associated with packet processing.

In NIC flow switching

Methods, apparatus, and systems for facilitating effective power management through dynamic reconfiguration of interrupts. Interrupt vectors are mapped to various processor cores in a multi-core processor, and interrupt workloads on the processor cores are monitored. When an interrupt workload for a given processor core is detected to fall below a threshold, the interrupt vectors are dynamically reconfigured by remapping interrupt vectors that are currently mapped to the processor core to at least one other processor core, such that there are no interrupt vectors mapped to the processor core after reconfiguration. The core is then enabled to be put in a deeper idle state. Similar operations can be applied to additional processor cores, effecting a collapsing of interrupt vectors onto fewer processor cores. In response to detecting cores emerging from idle states, reconfiguration of interrupt vectors can be performed to rebalance the assignment of the vectors across active cores by remapping a portion of the vectors to those cores.

Dynamic interrupt reconfiguration for effective power management

Scalable multi-tenant networking can preserve segregation of traffic of different tenants across multiple VLANs over a cluster of nodes. A single process is bound to a wildcard address and a port to listen for traffic across the VLANs and the cluster of nodes. The process detects receipt of a request at one of the multiple addresses assigned to the multiple VLANs and resolves the address to a name of a corresponding one of the VLANs. This is then used to determine an address of a node that is part of the VLAN and that hosts a volume identified in the received request. The requesting client is then redirected to the node that is part of the VLAN and that hosts the identified volume.

Scalable multiple VLAN multi-tenant networking

Data traffic of different customers or tenants can be efficiently handled at a shared node while still being isolated from each other. An application instance can create multiple network stack that are isolated from each other and intelligently manage threads across the isolated network stack instances. To intelligently manage the threads across the network stack instances, each thread maintains data that identifies the network stack to which the thread is assigned. With this information, the application can intelligently use a thread already assigned to a network stack that will process the data traffic and avoid the performance impact of a system call to assign the thread to the network stack.

Intelligent thread management across isolated network stacks

A method includes binding, using a plurality of processors, a process to a wildcard address and a port on each of a plurality of nodes. The process receives, on a redirector node, a first request for a first address of a first volume located on the cluster from a first client. The first request is sent to the port and a first address associated with a first virtual local area network (VLAN) that is not the wildcard address. The process determines the first address from the first request and a name of the first VLAN based on the first address. The process determines a first node that contains information regarding the first volume and an address of the first node that is part of the first VLAN. The process determines that a volume identifier associated with the first volume of the first request is present on a volume list.

Secure mode VLANs systems and methods

Described are embodiments of mediums, methods, and systems for application-reserved use of cache for direct I/O. A method for using application-reserved cache may include reserving, by one of a plurality of cores of a processor, use of a first portion of one of a plurality of levels of cache for an application executed by the one of the plurality of cores, and transferring, by the one of the plurality of cores, data associated with the application from an input/output (I/O) device of a computing device directly to the first portion of the one of the plurality of levels of the cache. Other embodiments may be described and claimed.

Application-reserved cache for direct I/O

Methods, systems, and apparatuses, including computer programs encoded on computer-readable media are disclosed for binding a process to a wildcard address and a port on a plurality of nodes of a cluster. A process receives a first request for a first address of a first volume that is sent to the port and a first address associated with a first virtual local area network that is not the wildcard address. The process determines the first address, a name of the first VLAN, and a first node that contains information regarding the first volume. The process determines an address of the first node that is part of the first VLAN. The address of the first node is returned. The process receives another request for a second address of a second volume that is sent to the port and a second address associated with another VLAN that is not the wildcard address.

Growing community of inventors

Portland, OR, United States of America

Peter P Waskiewicz, Jr