Sun Microsystems, Inc.

Oracle International Corporation

Oracle America, Inc.

Open Computing Trust I

Shimon Muller

Ariel Hendel

Arvind Srinivasan

Michael K Wong

Bjorn Dag Johnsen

Yatin Gajjar

Denton E Gentry, Jr

Rahoul Puri

Louise Y Yeung

Curt Berg

Howard M Frazier

Linda T Cheng

Ashok V Krishnamoorthy

Xuezhe Zheng

John E Watkins

Leick D Robinson

Binh Pham

David R Cheriton

Aly E Orady

Andreas V Bechtolsheim

Bjørn Dag Johnsen

Mohammad Kamel Issa

Ajoy Siddabathuni

William T Zaumen

Leo A Hejza

May Lin

Shree Murthy

Raju Penumatcha

Ravi Tangirala

Louise Yeung

Leo Hejza

System and method for aggressive credit waiting in a high performance computing environment. In accordance with an embodiment, systems and methods can provide for an indexed matrix of credit wait policies between ports within a single switch. In addition, systems and methods can provide for an array of credit wait polices at an egress port from a switch, the array being indexed by virtual lane.

System and method for supporting aggressive credit waiting in a high performance computing environment

The disclosed embodiments provide an optically switched network system. This system includes a passive optical switch with N inputs and N outputs, which can communicate different wavelengths from each of the N inputs to each of the N outputs. It also includes N end-nodes, and N pairs of optical fibers, wherein each pair connects one of the N end-nodes to one of the N inputs and one of the N outputs. The optically switched network is organized into a virtual data plane and a virtual control plane, which both communicate through the same underlying physical network. The virtual data plane provides any-to-all parallel connectivity for data transmissions among the N end-nodes. The virtual control plane is organized as a ring that serially connects the N end-nodes, wherein the ring communicates arbitration information among distributed-arbitration logic at each of the N end-nodes.

Optically switched network topology

System and method for supporting subnet number aliasing in a high performance computing environment. In accordance with an embodiment, a fabric member can be assigned, by a global fabric manager, an alias fabric local subnet number in order to keep a fabric running after a fabric reconfiguration. The alias fabric local subnet number can be assigned for a period of time, the period of time being static, configurable, or indefinite.

System and method for supporting subnet number aliasing in a high performance computing environment

System and method for supporting intra- and inter-subnet address resolution in a network environment using the same linear forwarding tale (LFT) for both the intra- and inter-subnet forwarding. Subnet prefix values in global route headers (GRHs) are used for linear forwarding table (LFT) lookup in a high performance computing environments. An exemplary can provide for use of an Inter Subnet Route Number (ISRN) embedded in the subnet prefix values in the GRHs for LFT lookup in a network switch environment in a high performance computing environment such as a network having an InfiniBand (IB) architecture. A method can provide, at a computer environment, including a network fabric, one or more subnets, each of which subnets are associated with one or more network switches or hosts. The system and method is compatible with legacy switches and nodes that are not conversant with the ISRNs.

System and method for using subnet prefix values in global route header (GRH) for linear forwarding table (LFT) lookup in a high performance computing environment

System and method for supporting multiple concurrent SL to VL mappings in a high performance computing environment. In accordance with an embodiment, systems and methods can provide for two or more SL to VL mappings per ingress switch port in a network switched fabric. By allowing for multiple such mappings, greater virtual lane independence can be achieved while continuing to achieve quality of service guarantees.

System and method for supporting multiple concurrent SL to VL mappings in a high performance computing environment

We disclose a method for controlling access to an optically switched network, which connects N end-nodes, and is organized into a virtual data plane and a virtual control plane, which both communicate through the same underlying physical optical network. The virtual data plane provides any-to-all parallel connectivity for data transmissions among the N end-nodes, and the virtual control plane is organized as a ring that serially connects the N end-nodes, wherein a control token circulates around the ring. During operation, an end-node in the ring receives the control token, which includes a destination-busy vector with a busy flag for each of the N end-nodes. If the end-node has data to send and the busy flag for the destination end-node is not set, the system: sets the busy flag; commences sending the data to the destination end-node; and forwards the control token to a next end-node in the ring.

Medium-access control technique for optically switched networks

System and method for supporting a partitioned switch forwarding table in a high performance computing environment. Described methods and systems can support partitioned switch forwarding tables (e.g., partitioned LFTs) by setting up hardware registers that divide the LFT into at least two partitions, a first partition that supports legacy forwarding (e.g., standard LID based forwarding without the need to use portions of the GRH), and a second partition to support the GRH based forwarding that is described above. In such a manner, switches and other hardware within a core fabric can behave as legacy nodes/switches having standard LFTs, while also being able to support the extended addressing supplied through the use of portions of the GRH.

System and method for supporting partitioned switch forwarding tables in a high performance computing environment

System and method for supporting flexible forwarding domain boundaries in a high performance computing environment. In accordance with an embodiment, flexible forwarding domain boundaries can be supported by dividing/partitioning a physical switch into two or more logical switches, where each switch is logically in a different domain, and allowing a fabric to be decomposed into independent subnets with one two or more physical end ports at the physical switch. By doing so, the same hierarchical forwarding structure and management structure between subnets can be provided as when complete physical switches are used as building blocks.

System and method for supporting flexible forwarding domain boundaries in a high performance computing environment

A system for communicating a multi-destination packet through a network switch fabric with a plurality of input and output ports is described. This system receives the multi-destination packet at an input port, wherein the multi-destination packet includes a multicast packet or a broadcast packet that is directed to multiple output ports, and wherein the network switch fabric maintains a separate virtual output queue (VOQ) for each output port. Next, the system sends the multi-destination packet from the input port to the multiple output ports by inserting the multi-destination packet into VOQs associated with the multiple output ports. The multi-destination packet is inserted into one VOQ at a time, so that after the multi-destination packet is read out of a VOQ and is sent to a corresponding output port, the multi-destination packet is inserted in another VOQ until the multi-destination packet is sent to all of the multiple output ports.

Scheduling packets with multiple destinations in a virtual output queue network switch

An optically switched network system includes an optical switch with N inputs and N outputs that connects N end-nodes and is structured to transmit N wavelengths from each of the N inputs to each of the N outputs. The system includes a virtual data plane and a virtual control plane, which both communicate through the optical switch. The virtual data plane provides any-to-all parallel connectivity for data transmissions among the N end-nodes. The N end-nodes are partitioned into two or more subsets, wherein end-nodes in a given source subset transmit data to a given destination subset using wavelengths, which are not used by end-nodes outside of the given source subset to transmit data to the same given destination subset. The virtual control plane includes two or more rings associated with the two or more subsets of end-nodes. Each ring passes through a subset of end-nodes, and is used to communicate arbitration information among arbitration logic located at each end-node in the ring.

Scalable, low-latency medium-access control technique for optically switched networks

Upon detecting a failure of a switch link in a set of multiple redundant switch links that directly connect two nodes in the switch fabric network, the system performs a failover operation that remaps network traffic associated with the failed switch link to an alternative switch link in the set of multiple redundant switch links. Each node in the switch fabric network maintains a mapping table that translates from logical ports and associated logical virtual lanes to physical ports and associated virtual lanes. The system also provisions switch links with private virtual lanes comprising resources to facilitate failover operations. While remapping the network traffic, the system changes the mapping table so that network traffic, which is presently directed to a physical port and associated virtual lane for the failed switch link, is remapped to an alternative physical port and an associated private virtual lane, for the alternative switch link.

Scalable framework for managing QoS requirements after a switch link failover in a switch fabric network

The disclosed system handles a switch link failure in a switch fabric network. When a node in the switch fabric network detects a failure of a switch link coupled to the node, the system remaps traffic that is directed to a logical port number, which is currently mapped to a primary physical port number associated with the failed switch link, to a secondary physical port number associated with the alternative switch link. This remapping involves performing a lookup in a local mapping table at the node, wherein the local mapping table stores associations between logical port numbers and physical port numbers, wherein for each logical port number, the mapping table includes a primary physical port number and one or more secondary physical port numbers, which are associated with alternative switch links. The system notifies a subnet manager for the switch fabric network about the link failure and the remapping.

Growing community of inventors

Sunnyvale, CA, United States of America

Shimon Muller