Cisco Technology, Inc.

International Business Machines Corporation

Lenovo (singapore) Pte. Ltd.

A9.com, Inc.

Santosh Rajagopalan

Hariharan Balasubramanian

Ronak K Desai

Sanjay Sane

Jayakrishna Kidambi

Keshav G Kamble

Dar-Ren Leu

Amitabha Biswas

Ayan Banerjee

Chandan Mishra

Leonard Thomas Tracy

Uday Shankar Nagaraj

Smita Rai

Gayatri Ramachandran

Ajay Kulhari

Tang Gang

Vijoy A Pandey

Vinit Jain

Abhay Kumar Roy

Stephan Benny

Nilanjan Mukherjee

Raghava Kodigenahalli Sivaramu

Varun A Shah

Christian Sasso

Ankur Goyal

Tameen Khan

Narayanan Sadagopan

Trishul Amit Madhukar Chilimbi

Vishwanathan Swaminathan

Hyokun Yun

Vithal Shirodkar

Sankalp Nayak

Choon Hui Teo

Iman Keivanloo

Devices and techniques are generally described for search using a micro-partitioned catalog. In some examples, a search system can determine a plurality of items in an online catalog. The search system can generate a similarity graph based on vector representations of the plurality of items and determine an item as a representative item for a cluster of items. The search system can receive a query and determine the cluster of items as a candidate cluster based on a comparison between the query and the representative item. The search system can determine a candidate item of the cluster of items based on a comparison between the query and the cluster of items.

Micro-partitioning based search

Consistent with embodiments of the present disclosure, systems and methods are disclosed for providing per-group ECMP for multidestination traffic in a DCE/TRILL network. Embodiments enable per-group load balancing of multidestination traffic in DCE/L2MP networks by creating a new IS-IS PDU to convey the affinity of the parent node for a given multicast group. For broadcast and unknown unicast flooded traffic, the load balancing may be done on a per-vlan basis.

ECMP parent group selection for multidestination traffic in DCE/TRILL networks

In an embodiment, a method comprises: detecting a change in a multiple-switch configuration in a data communications network comprising a plurality of packet data switches configured as roots of multicast trees. In response to detecting that the multiple-switch configuration has changed, a first value, a second value and a third value representing limits on a number of multicast trees supported in the network and prioritization of the switches are retrieved. The method further comprises determining a type of the multiple-switch configuration change. In response to determining that the type indicates that a first switch was added to the multiple-switch configuration, using at least the first, second and third values, it is determined whether to configure the first switch as a first root in the multiple-switch configuration. The method is performed by one or more computing devices.

Determining multicast root nodes in a telecommunications network

In one embodiment, a server includes a processor and logic integrated with and/or executable by the processor. The logic is configured to hash at least a portion of a packet according to a hashing algorithm to obtain a hash value. The logic is also configured to select an uplink with which to forward the packet based on the hash value. In another embodiment, a method includes hashing at least a portion of at least one packet according to a hashing algorithm to obtain a hash value using an accelerated network interface card (NIC) of a server. The accelerated NIC is configured to provide overlay functionality. The method also includes selecting an uplink, based on the hash value, from a plurality of uplinks available to send traffic out to a network using the accelerated NIC.

Load balancing overlay network traffic using a teamed set of network interface cards

In one embodiment, a method includes splitting a plurality of physical ports distributed across multiple distributed line cards (DLCs) into a plurality of virtual switches, wherein each virtual switch comprises ports of one or more DLC, creating a virtual control domain (VCD) associated with each virtual switch, and autonomously controlling the virtual switches, wherein each DLC is in electrical communication with at least one switch fabric coupler (SFC), and wherein the plurality of physical ports appear to external devices as being part of a single cell switched domain. Other systems and methods are described according to more embodiments.

Method for controlling large distributed fabric-based switch using virtual switches and virtual controllers

In one embodiment, a system includes at least one distributed line card (DLC) in electrical communication with at least one switch fabric coupler (SFC) and a meta-controller in electrical communication with the at least one DLC, the meta-controller including a processor for executing logic, logic configured for creating more than one virtual switch and a corresponding virtual control domain (VCD) associated with each virtual switch, wherein each virtual switch includes a plurality of physical ports of one or more of the at least one DLC, and logic configured for autonomously controlling the virtual switches, wherein the system appears to a device external of the system as a single cell switched domain. Other systems are described according to more embodiments.

Large distributed fabric-based switch using virtual switches and virtual controllers

A system includes a server including: logic adapted for receiving traffic from a virtual machine (VM), the traffic including at least one packet, logic adapted for hashing at least a portion of the at least one packet according to a hashing algorithm to obtain a hash value, and logic adapted for selecting an uplink based on the hash value; at least one accelerated network interface card (NIC), each accelerated NIC including: network ports including multiple Peripheral Component Interconnect express (PCIe) ports adapted for communicating with the server and a network, each network port including an uplink, logic adapted for encapsulating the at least one packet into an overlay-encapsulated packet, logic adapted for storing a media access control (MAC) address corresponding to the selected uplink as a source MAC (SMAC) address in an outer header of the overlay-encapsulated packet, and logic adapted for sending the overlay-encapsulated packet via the selected uplink.

In one embodiment, a system includes a server running a virtualization platform, the virtualization platform including logic adapted for creating one or more virtual machines (VMs) and logic adapted for managing a virtual switch (vSwitch), a controller in communication with the server, the controller including logic adapted for assigning a media access control (MAC) address and a virtual local area network (VLAN) identifier (ID) to each of the one or more VMs, wherein a specific tenant to which the one or more VMs belongs is indicated using a tenant ID derived from the VLAN ID, the MAC address, or a combination thereof. Other systems, methods, and computer program products are also described according to more embodiments.

Scalable network overlay virtualization using conventional virtual switches

An example method is provided and includes a multicast data message from a data source, the message in a first virtual local area network and being associated with a multicast group. The method also includes calculating a hash value based on the virtual local area network, the data source, and the multicast group, determining a port for a designated router in a Layer-2 network based on the hash value, and switching the multicast data message to the port that was determined.

System and method for layer-2 multicast multipathing

Techniques are described which facilitate multi-destination forwarding in a Layer 2 Multipath (L2MP) network which includes an emulated switch. The emulated switch may correspond to two or more underlying peer link switches in the L2MP network, in which each of the peer link switches is linked to a Classical Ethernet (CE) switch over a virtual port channel (vPC). Traffic received by one of the peer link switches over the vPC is automatically forwarded to the other peer link switch (or switches). Multi-destination frames originating from the L2MP network addressed to hosts within the CE network are sent over only one of the peer link switches.

Multi-destination forwarding in network clouds which include emulated switches

In one embodiment, a method includes establishing a link between two N_Port Identifier Virtualization (NPIV) switches, the link having a high cost assigned thereto. The NPIV switches are in communication with a plurality of hosts through an N_Port Virtualization (NPV) device. The method further includes receiving at a first of the NPIV switches, an indication of a failure at a second of the NPIV switches, receiving data at the first NPIV switch, the data destined for one of the hosts associated with a domain of the second NPIV switch, and forwarding the data to the NPV device for delivery to the host, wherein a Fibre Channel Identifier (FCID) of the host is the same before and after the failure at the second NPIV switch. An apparatus is also disclosed.

[object Object]

In one embodiment, a method includes obtaining an indication that a state associated with a node is to be changed and preventing data from being received on a first link. The method also includes updating at least one selected from a group including an incoming interface check (IIC) table and an outgoing interface (OIF) table to reflect the state. The state indicates that a second link is to be activated. Finally, the method includes allowing the data to be received on the second link after updating either or both the IIC table and the OIF table and after the timer duration for the NULL value for IIC has expired.

Method and apparatus for sequencing operations for an incoming interface check in data center ethernet

In one embodiment, a method includes assigning at a switch in a layer 2 multipath network, costs to a link in the network, each of the link costs associated with a different graph for forwarding traffic in the network, transmitting the link costs to other switches in the layer 2 multipath network, and receiving and forwarding traffic according to one of the graphs. An apparatus is also disclosed.

Per-graph link cost assignment in layer 2 multipath networks

In one embodiment, a method includes receiving at a processor at a node, notification of an error in a VLAN to topology mapping at the node, receiving a multi-destination packet from the VLAN in the VLAN to topology mapping, the multi-destination packet including a tree identifier associated with one of the topologies, and transmitting the multi-destination packet to all forwarding ports at the node in an unpruned tree corresponding to the tree identifier contained in the multi-destination packet. An apparatus is also disclosed.

Growing community of inventors

San Jose, CA, United States of America

Santosh Rajagopalan