Hewlett-packard Development Company, L.p.

Hewlett Packard Enterprise Development LP

Digital Equipment Corporation

Google Inc.

Other

Compaq Computer Corportion

Jeffrey Clifford Mogul

Praveen Yalagandula

Jayaram Mudigonda

Parthasarathy Ranganathan

Marcos Kawazoe Aguilera

Sujata Banerjee

Yuhong Mao

Dwight L Barron

Paul T Congdon

Lucian Popa

Puneet Sharma

Benjamin Charles Serebrin

Keith Farkas

Mehul A Shah

Yoshio Frank Turner

Jean Tourrilhes

Janet Lynn Wiener

Jose Renato G Santos

Jakov Seizovic

Kevin B Leigh

Terrel L Morris

Michael Burrows

Richard Lee Sites

Paolo Faraboschi

Michael S Schlansker

Naveen Muralimanohar

Andrei Z Broder

Nathan Lorenzo Binkert

Jung Gun Lee

Eric A Anderson

Kevin T Lim

Raymond Paul Stata

Richard Fitzhugh Wrenn

Michael David Mitzenmacher

Bryan E Stiekes

Mallikarjun Chadalapaka

Guodong Zhang

Randolph B Haagens

Anna Fischer

Alvin Auyoung

Eduardo Argollo De Oliveira Dias, Jr

Richard J Wagner

Eduardo S Pinheiro

Lawrence S Brakmo

Patrick A Reynolds

Jayaram Mudlgonda

Bryan Stiekes

Disclosed herein are a system, non-transitory computer readable medium, and method for governing communications of a bare metal guest in a cloud network. A network interface handles packets of data in accordance with commands by a control agent.

Governing bare metal guests

The present application describes a system and method for rate limiting traffic of network users, such as virtual machines (VMs). In accordance with the disclosure, transmission queues for the VMs may be assigned to two levels of rate limiting. The first-level rate limit may be an individual rate limit for a particular transmission queue, while the second-level rate limit may be a group rate limit that is applied to a group of transmission queues. The first-level rate limit for some transmission queues may be performed using software rate limiting, while for other transmission queues the first-level rate limit may be performed using hardware rate limiting. The second-level rate limit may be applied to a group of transmission queues that contains both software first-level rate limited transmission queues and hardware first-level rate limited transmission queues.

Network multi-level rate limiter

A network switch for implementing a Software Defined Network can include one or more memory resources storing a flow table and including a buffer. The network switch can receive and analyze data packets to generate a lookup key for each respective data packet. Using the lookup key, the network switch can perform a lookup in the flow table to determine a rule number for the respective data packet. Based on the rule number, the network switch can generate an event record for the respective data packet, and store the event record in a block of the buffer. When the block is full, the network switch can stream a set of event records from the buffer to a processing resource to enable the processing resource to update counter tables, external to the network switch, to implement the Software Defined Network.

Implementing a software defined network based on event records

An example method for allocating resources in accordance with aspects of the present disclosure includes collecting proposals from a plurality of modules, the proposals assigning the resources to the plurality of modules and resulting in topology changes in a computer network environment, identifying a set of proposals in the proposals, the set of proposals complying with policies associated with the plurality of modules, instructing the plurality of modules to evaluate the set of proposals, selecting a proposal from the set of proposals, and instructing at least one module associated with the selected proposal to instantiate the selected proposal.

Resource allocator

Selected ones of multiple groups of network entities are assigned to corresponding routing structures, where each of the routing structures defines a set of paths through a network among network entities of a corresponding one of the groups. The assigning considers available multicast resources in communication nodes in the network, and a criterion relating to reducing overall cost in the assignment of the selected groups to the corresponding routing structures.

Assigning selected groups to routing structures

According to an example, a method for bandwidth guarantee and work conservation includes determining virtual machine (VM) bandwidth guarantees assigned to VMs in a network including a source VM that communicates with destination VMs. The method further includes assigning minimum bandwidth guarantees to communications between the source VM with the destination VMs by dividing a VM bandwidth guarantee assigned to the source VM between the destination VMs based on active VM-to-VM communications between the source VM and the destination VMs. The method also includes allocating, by a processor, spare bandwidth capacity in the network to a communication between the source VM and a destination VM based on the assigned minimum bandwidth guarantees.

Bandwidth guarantee and work conservation

Example methods, apparatus, and articles of manufacture to control generation of memory access requests in processor systems are disclosed. A disclosed example method involves determining at a memory controller whether a memory access queue depth for a memory reference is greater than a first threshold. When the memory access queue depth for the memory reference is greater than the first threshold, an indication is sent to control generation of memory access requests corresponding to the memory reference.

Methods and apparatus to control generation of memory access requests

The present application describes a system and method for a virtual machine to classify a packet. Once the virtual machine (VM) classifies the packet, it bypasses a hypervisor to enqueue the packet directly on a hardware transmission queue. The NIC will then verify that the VM classified and enqueued the packet correctly. If the packet was classified properly, it is transmitted over the wire to its destination. In this regard, the system and method provides a technique for verifying that the VM is enqueuing packets properly, while improving performance by allowing high-rate flows to bypass the hypervisor.

Method and system for validating rate-limiter determination made by untrusted software

The present application describes a system and method for rate limiting traffic of a virtual machine (VM). In this regard, a VM bypasses a hypervisor and enqueues a packet on an assigned transmission queue. Based on information contained in the packet, the NIC determines whether the packet is to be delayed or transmitted immediately. If the NIC determines that the packet is to be transmitted immediately, the packet is moved to one of a plurality of primary output queues to be transmitted to the external network. If the packet is to be delayed, the packet is moved to one of a plurality of rate limited secondary output queues. In this regard, the NIC classifies the packets, thereby improving performance by allowing high-rate flows to bypass the hypervisor.

Method and system for enforcing multiple rate limits with limited on-chip buffering

A system, method, and non-transitory computer-readable medium for filtering flow table entries is described. A trusted device may send a table read request to a network interface controller (NIC) that includes predicate logic. The NIC will use the received predicate logic to filter flow table entries. The entries that satisfy the criteria specified by the trusted software are moved to a Direct Memory Access (DMA) buffer on the NIC to create a subset of entries. Once some or all of the flow table has been scanned, the subset of entries may be transferred from the DMA buffer on the NIC to a memory of the trusted host via a DMA transfer to be further processed by the trusted host.

Method and system for filtering flow table counters

A system and method is provided for generating network timeouts by utilizing intentionally delayed message transmissions sent internal to a network controller. For example, according to this system and method a network controller transmits a data message externally over a network and the data message invokes a response network message. The network controller also transmits a timeout message corresponding to the data message; however, the timeout message is intentionally delayed for a predetermined duration of time prior to being internally transmitted. controller host device receives at least one of the response message and the timeout message and determines which is received first. If the timeout message is received, a request associated with the transmitted data message is treated as having timed out.

Network timeouts using intentionally delayed transmissions

A network switch for reducing TCP timeouts due to incast collapse is disclosed. The network switch has a plurality of buffers for storing network packets for at least one sender server. An incast buffer management component in the network switch allocates a buffer in the plurality of buffers to store an incast notification packet, the incast notification packet indicating one or more network packets that have been dropped by the network switch. The network switch also includes an incast notification component for forwarding the incast notification packet to a destination server with an incast notification packet specific header.

Reducing TCP timeouts due to Incast collapse at a network switch

A method of generating a plurality of potential generalized fat tree network (GFT) topologies is provided herein. The method includes receiving parameters that specify a number of servers, a number of switches, and a number of ports in the switches. The parameters are for configuring an GFT topology. The method also includes generating a plurality of potential GFT topologies based on heuristics that favor higher bandwidth and lower latency. Generating the plurality of potential GFT topologies includes generating each level of the potential GFT topologies, recursively, and from a bottom level of a potential GFT topology to a top level.

Generating EGFT network topologies

Each of a plurality of data packets that is received on a network switch can be associated with a corresponding event record that identifies one or more flow table rules for that data packet. A plurality of event records, which individually can correspond to one or more of the data packets, is transmitted to a programmable resource for use in controlling a Software Defined Network.

Implementing a software defined network using event records that are transmitted from a network switch

A scalable, multi-tenant network architecture for a virtualized datacenter is provided. The network architecture includes a network having a plurality of servers connected to a plurality of switches. The plurality of servers hosts a plurality of virtual interfaces for a plurality of tenants. A configuration repository is connected to the network and each server in the plurality of servers has a network agent hosted therein. The network agent encapsulates packets for transmission across the network from a source virtual interface to a destination virtual interface in the plurality of virtual interfaces for a tenant in the plurality of tenants. The packets are encapsulated with information identifying and locating the destination virtual interface, and the information is interpreted by switches connected to the source virtual interface and the destination virtual interface.

Scalable multi-tenant network architecture for virtualized datacenters

Growing community of inventors

Menlo Park, CA, United States of America

Jeffrey Clifford Mogul