Arrcus Inc.

Kalyani Rajaraman

Keyur Patel

Nalinaksh Pai

Vikram Ragukumar

Lalit Kumar

Derek Man-Kit Yeung

Neeraj Malhotra

Robert Austein

Sridhar Pitchai

Ebben Aries

Randall Bush

Rajkumar Gurusamy

Ashutosh Shanker

Tetsuya Murakami

Ravindra Vaishampayan

Matthew Anderson

Lawrence Rolfe Kreeger

Shitanshu Shah

Systems, methods, and devices for offloading network data to a datastore. A system includes routing chip hardware and an asynchronous object manager in communication with the routing chip hardware. The asynchronous object manager is configurable to execute instructions stored in non-transitory computer readable storage media. The instructions include asynchronously receiving a plurality of objects from one or more producers. The instructions include identifying one or more dependencies between two or more of the plurality of objects. The instructions include reordering the plurality of objects according to the one or more dependencies. The instructions include determining whether the one or more dependencies is resolve. The instructions include, in response to determining the one or more dependencies is resolved, calling back an application and providing one or more of the plurality of objects to the application.

Asynchronous object manager in a network routing environment

Systems, methods, and devices for improved routing operations in a network computing environment. A system includes a first switch and a second switch in a network topology. The system includes a host virtual machine in communication with at least one of the first switch and the second switch. The system includes a routed peer link connecting the first switch to the second switch. The system is such that the first switch and the second switch have the same Internet protocol (IP) address and media access control (MAC) address.

First hop gateway redundancy in a network computing environment

Systems, methods, and devices for improved routing operations in a network computing environment. A system includes a network topology comprising a plurality of spine nodes and a plurality of leaf nodes, wherein a link between a first spine node and a first leaf node is inactive. The first spine node includes one or more processors configurable to execute instructions stored in non-transitory computer readable storage media. The instructions include receiving a packet to be transmitted to the first leaf node. The instructions include identifying an alternative spine node at a same level in the network topology. The instructions include attaching a tunnel label to the packet, wherein the tunnel label indicates the packet should be transmitted to the alternative spine node.

Loop conflict avoidance in a network computing environment

Systems, methods, and devices for improved routing operations in a network computing environment. A system includes a network topology comprising a spine node and a plurality of leaf nodes. The system is such that at least one of the plurality of leaf nodes is associated with one or more networking prefixes. The spine node stores a prefix table. The prefix table includes a listing of networking prefixes in the network topology. The prefix table includes an indication of at least one equal-cost multipath routing (ECMP) group associated with each of the networking prefixes in the network topology. The prefix table includes an indication of at least one leaf node of the plurality of leaf nodes associated with each of the networking prefixes in the network topology.

Routing optimizations in a network computing environment

A segment identifier (SID) manager interfaces with clients by way of a library. Clients implementing network services and needing SIDs according to internet SRv6, are allocated blocks by the SID manager using the library. Clients assign SIDs from the block independent of the SID manager using the library. Where a threshold portion of the block is assigned, the library requests an additional block of SIDs from the SID manager. Clients transmit context mappings of assigned SIDs to the SID manager for use by administrators and to enable recovery in the event of failure.

SID allocation

A logical router includes disaggregated network elements that function as a single router and that are not coupled to a common backplane. The logical router includes spine elements and leaf elements implementing a network fabric with front panel ports being defined by leaf elements. Control plane elements program the spine units and leaf to function a logical router. The control plane may define operating system interfaces mapped to front panel ports of the leaf elements and referenced by tags associated with packets traversing the logical router. Redundancy and checkpoints may be implemented for a route database implemented by the control plane elements. The logical router may include a standalone fabric and may implement label tables that are used to label packets according to egress port and path through the fabric.

Logical router comprising disaggregated network elements

Disaggregated border gateway protocol (BGP) enables an eBGP session between an internal node an external node to continue despite failover of a perimeter through which the eBGP session is established. eBGP control traffic is trapped by a perimeter router and forwarded to a BGP speaker on the internal node through an IP tunnel. Failover is detected in response to a change in a source address of the IP tunnel over which eBGP control traffic is received. The BGP speaker announces routes to the external node that include a reference to an internal address of an active perimeter router. In response to failover, the BGP speaker announces updated routes referencing the standby router for the perimeter router.

Disaggregated border gateway protocol (BGP)

Systems, methods, and devices for routing operations in a network computing environment. A system includes a network topology comprising a plurality of spine nodes and a plurality of leaf nodes, wherein a link between a first spine node and a first leaf node is inactive. The first spine node includes one or more processors configurable to execute instructions stored in non-transitory computer readable storage media. The instructions include receiving a packet to be transmitted to the first leaf node. The instructions include identifying an alternative spine node at a same level in the network topology. The instructions include attaching a tunnel label to the packet, wherein the tunnel label indicates the packet should be transmitted to the alternative spine node.

Growing community of inventors

San Jose, CA, United States of America

Kalyani Rajaraman