Abstract: A system comprises an edge services controller configured to: compute, based on a physical topology of physical links that connect a plurality of network interface cards (NICs) that comprise embedded switches and processing units coupled to the embedded switches, a virtual topology comprising a strict subset of the physical links; and program the virtual topology into the respective processing units of the NICs to cause the processing units of the NICs to send data packets via physical links in the strict subset of the physical links.

Abstract: In general, techniques are described for automatically and transparently providing service proxying to virtual machines using Network Interface Cards (NICs). In some examples, a service proxy implemented by a NIC of a computing device that hosts a virtual machine may mimic sidecar service proxy behavior. A NIC-based mesh agent may automatically detect the service offered by the virtual machine and interact with control plane components to dynamically incorporate the service into a service mesh.

Abstract: Techniques are described for specifying a backend virtual network for a service load balancer. An example orchestrator of this disclosure is configured to receive a service definition for a service implemented by load balancing service traffic for the service among a plurality of backend virtual execution elements, wherein the service definition specifies a first virtual network to use as a backend virtual network for the service, to instantiate, in a selected one of the computing devices, a backend virtual execution element for the service, and to configure, based on the service definition specifying the first virtual network to use as the backend virtual network for the service, a network controller for the virtualized computing infrastructure to configure a load balancer to load balance service traffic to a first virtual network interface, of the backend virtual element, for the first virtual network.

Abstract: In general, techniques are described for a computing device including a virtual router, a pod comprising a container, and a network plugin. The virtual router includes a virtual router agent. The network plugin includes processing circuitry configured to receive, from the virtual router agent, an indication of an interface type for a virtual network for the pod and to configure, for the pod, a virtual network interface having the interface type, the virtual network interface for communicating on the virtual network.

Abstract: Techniques are described for specifying a backend virtual network for a service load balancer. An example orchestrator of this disclosure is configured to receive a service definition for a service implemented by load balancing service traffic for the service among a plurality of backend virtual execution elements, wherein the service definition specifies a first virtual network to use as a backend virtual network for the service, to instantiate, in a selected one of the computing devices, a backend virtual execution element for the service, and to configure, based on the service definition specifying the first virtual network to use as the backend virtual network for the service, a network controller for the virtualized computing infrastructure to configure a load balancer to load balance service traffic to a first virtual network interface, of the backend virtual element, for the first virtual network.

Abstract: Techniques are described for performing latency-aware load balancing. In some examples, a computing device communicably coupled to a plurality of service endpoints that are in motion with respect to the computing device may receive data to be processed. The computing device may select, based at least in part on a communication latency of each of the plurality of service endpoints and a predicted compute latency of each of the plurality of service endpoints, a service endpoint out of the plurality of service endpoints to process the data. The computing device may send the data to the selected service endpoint for processing.

Abstract: In general, techniques are described for a computing device including a virtual router, a pod comprising a container, and a network plugin. The virtual router includes a virtual router agent. The network plugin includes processing circuitry configured to receive, from the virtual router agent, an indication of an interface type for a virtual network for the pod and to configure, for the pod, a virtual network interface having the interface type, the virtual network interface for communicating on the virtual network.

Abstract: Techniques are described for learning an unknown virtual network information, such as an virtual Internet Protocol (IP) address, of a pod in a virtual network. In some examples, a virtual router executing at a computing device may receive an Address Resolution Protocol (ARP) packet from a virtual execution element in the virtual network, the virtual execution element executing at the computing device. The virtual router may determine, based at least in part on the ARP packet, whether virtual network information for the virtual execution element in a virtual network is known to the virtual router. The virtual router may, in response to determining that the virtual network information of the virtual execution element in the virtual network is not known to the virtual router, perform learning of the virtual network information for the virtual execution element.

Abstract: Techniques are described for creating isolated pools of external, failover, and/or floating IP addresses. In one example, this disclosure describes a method including creating a plurality of virtual networks, creating a plurality of pools of external IP addresses, detecting a request to instantiate an object that identifies a specific pool from the plurality of pools of external IP addresses; and instantiating the object and configuring the object with an external IP address drawn from the specific pool. The pools of external IP addresses may be created and isolated on a per-namespace, per-service, or per-ingress basis.

Abstract: A software defined networking (SDN) controller is configured to receive, from a Top-Of-Rack (TOR) switch, a first multicast route and a second multicast route. In response to determining that the first multicast route is an assisted replication route, the SDN controller is configured to add a first nexthop specified by the first multicast route to a list of nexthops for Broadcast, Unknown-Unicast, and Multicast (BUM) traffic. In response to determining that the second multicast route is not the assisted replication route, the SDN controller is configured to refrain from adding a second nexthop specified by the second multicast route to the list of nexthops. After adding the first nexthop, the SDN controller is configured to provision the list of nexthops at a virtual router.

Abstract: Techniques are described for performing latency-aware load balancing. In some examples, a computing device communicably coupled to a plurality of service endpoints that are in motion with respect to the computing device may receive data to be processed. The computing device may select, based at least in part on a communication latency of each of the plurality of service endpoints and a predicted compute latency of each of the plurality of service endpoints, a service endpoint out of the plurality of service endpoints to process the data. The computing device may send the data to the selected service endpoint for processing.

Abstract: Techniques are described for learning an unknown virtual network information, such as an virtual Internet Protocol (IP) address, of a pod in a virtual network. In some examples, a virtual router executing at a computing device may receive an Address Resolution Protocol (ARP) packet from a virtual execution element in the virtual network, the virtual execution element executing at the computing device. The virtual router may determine, based at least in part on the ARP packet, whether virtual network information for the virtual execution element in a virtual network is known to the virtual router. The virtual router may, in response to determining that the virtual network information of the virtual execution element in the virtual network is not known to the virtual router, perform learning of the virtual network information for the virtual execution element.

Abstract: A container orchestration platform manages a plurality of instances of resources including a first custom resource and a second custom resource. An API server of the container orchestration platform receives a request to delete an instance of the second custom resource; determines whether instance data associated with the instance of the second custom resource has a backreference identifying an instance of the first custom resource, the backreference indicating the instance of the first custom resource is dependent on the instance of the second custom resource; and in response to determining that the instance data has the backreference to the instance of the first custom resource, bypasses deletion of the instance of the second custom resource.

Abstract: An example system comprises a plurality of servers comprising respective network interface cards (NICs) connected by physical links in a physical topology, wherein each NIC of the plurality of NICs comprises an embedded switch and a processing unit coupled to the embedded switch; and an edge services controller configured to program the processing unit of a network interface card of the plurality of network interface cards to: receive, at a first network interface of the NIC, a data packet from a physical device; based on the data packet being received at the first network interface, modify the data packet to generate a modified data packet; and output the modified data packet to the physical device via a second network interface of the NIC.

Abstract: An example method comprises, receiving resource availability values from the plurality of Network Interface Cards (NICs); determining a data path for data packets of a flow transported using a protocol from a source NIC to a destination NIC via a NIC set that comprises at least one NIC, wherein: the plurality of NICs comprises the source NIC, the destination NIC, and the NIC set, and determining the data path comprises selecting the NIC set based on the resource availability values; and transmitting, to the source NIC and to each NIC in the NIC set, data path data to cause the source NIC and each NIC in the NIC set to identify the data packets of the flow using an identifier of the protocol and to transmit the data packets of the flow from the source NIC to the destination NIC via the data path.

Abstract: A system comprises an edge services controller configured to: compute, based on a physical topology of physical links that connect a plurality of network interface cards (NICs) that comprise embedded switches and processing units coupled to the embedded switches, a virtual topology comprising a strict subset of the physical links; and program the virtual topology into the respective processing units of the NICs to cause the processing units of the NICs to send data packets via physical links in the strict subset of the physical links.

Abstract: A system is configured to compute a latency between a first computing device and a second computing device. The system includes a network interface card (NIC) of a first computing device. The NIC includes a set of interfaces configured to receive one or more packets and send one or more packets. The processing unit is configured to identify information indicative of a forward packet, compute, based on a first time corresponding to the forward packet and a second time corresponding to a reverse packet associated with the forward packet, a latency between the first computing device and a second computing device, wherein the second computing device includes a destination of the forward packet and a source of the reverse packet, and output information indicative of the latency between the first computing device and the second computing device.

Abstract: Techniques are described for performing latency-aware load balancing. In some examples, a computing device communicably coupled to a plurality of service endpoints that are in motion with respect to the computing device may receive data to be processed. The computing device may select, based at least in part on a communication latency of each of the plurality of service endpoints and a predicted compute latency of each of the plurality of service endpoints, a service endpoint out of the plurality of service endpoints to process the data. The computing device may send the data to the selected service endpoint for processing.

Abstract: Example techniques and computing devices are disclosed. An example computing device includes a first non-uniform memory access (NUMA) node and a second NUMA nod. The first NUMA node includes a first network interface card, a first virtual router for one or more virtual networks, the first virtual router comprising first processing circuitry and configured with a first virtual host interface having a first Internet Protocol (IP) address, and a first workload executing on the first NUMA node. The second NUMA node includes a second network interface card, a second virtual router for the one or more virtual networks, the second virtual router comprising second processing circuitry and configured with a second virtual host interface having a second IP address, and a second workload executing on the second NUMA node.

Abstract: In general, the disclosure describes examples where a single software-defined network (SDN) controller is configured to receive an indication of a first cluster identifier for a first cluster of computing devices and receive an indication of a second cluster identifier for a second cluster of computing devices. In response to a determination that first configuration information indicates the first cluster identifier, the SDN controller is configured to configure a first set of virtual routers at the first cluster to connect the first group of workloads to a network using the first configuration information. In response to a determination that second configuration information indicates the second cluster identifier, the SDN controller is configured to configure a second set of virtual routers at the second cluster to connect the second group of workloads to a network using the second configuration information.