Abstract: Various embodiments set forth techniques for taking a snapshot of virtual memory of a virtual machine. One technique includes allocating, in a persistent memory, one or more blocks associated with a virtual memory; detecting a write request associated with a first portion of the virtual memory; in response to detecting the write request associated with the first portion, prioritizing the first portion; based on the prioritizing, copying the first portion into the one or more blocks in the persistent memory ahead of a second portion of the virtual memory; and after copying the first portion: applying the write request to the first portion; and copying the second portion into the one or more blocks in the persistent memory.

Abstract: Various embodiments set forth techniques for managing devices in a virtual environment. The techniques include mapping a virtual device of a virtual computing instance to a first device implementation. The techniques further include, in response to a trigger condition, remapping the virtual device to a second device implementation. The remapping is transparent to the virtual computing instance.

Abstract: Various embodiments disclosed herein are related to a non-transitory computer readable storage medium. In some embodiments, the medium includes instructions stored thereon that, when executed by a processor, cause the processor to send an indication of a first storage location to a destination host. In some embodiments, the first storage location includes content that is swapped out from a memory location in a source host. In some embodiments, the indication includes one or more of a logical address and a first physical address. In some embodiments, the medium includes instructions stored thereon that, when executed by a processor, cause the processor to map the logical address of the first storage location to a second physical address of a second storage location. In some embodiments, the destination host accesses the content of the first storage location.

Abstract: Techniques for migrating a virtual machine from a source node to a target node. A source node hosting a virtual machine is interconnected over a hardware-assisted interconnection fabric to a target node that is configured to receive all or portions of a to-be-migrated virtual machine. During migration, some portions of the virtual machine's memory contents might be at the source node, whereas other portions of the virtual machine's memory might be at the target node. When a CPU is processing a next instruction of a first virtual machine running on a source node, then rather than accessing physical memory of the source node, instead using the hardware-assisted interconnection fabric to fetch the next instruction from the target node. CPUs that are executing virtual machine code at either the source node or the target node can fetch instructions from memory at either node, or from memory of the hardware-assisted interconnection fabric.

Abstract: Methods, systems, and computer code for CPU-oblivious replication of memory of a virtual machine from one computing node to another computing node over PCIe hardware. A source virtual machine is designated to be replicated from a source computing node of a multi-node computing cluster to a target computing node of the multi-node computing cluster. When a CPU of the source computing node decodes an instruction that causes a change to contents of a memory location of the source virtual machine, the changed contents of the memory location is copied into a memory address of a target virtual machine by using hardware capabilities of two or more PCIe devices to synchronously clone data from the source computing node to the target computing node. The CPU of the source computing node obliviously waits to execute a further instruction of the source virtual machine until accomplishment of the cloned memory WRITE transaction.

Abstract: Various embodiments set forth techniques for taking a snapshot of virtual memory of a virtual machine. One technique includes allocating, in a persistent memory, one or more blocks associated with a virtual memory, annotating a first portion of the virtual memory for copying in a first pass, copying the first portion into the one or more blocks in the persistent memory in the first pass, receiving a write request associated with the first portion, and in response to receiving the write request: applying the write request to the first portion and annotating the first portion for copying in a second pass subsequent to the first pass.

Abstract: Various embodiments set forth techniques for taking a snapshot of virtual memory of a virtual machine. One technique includes allocating, in a persistent memory, one or more blocks associated with a virtual memory; detecting a write request associated with a first portion of the virtual memory; in response to detecting the write request associated with the first portion, prioritizing the first portion; based on the prioritizing, copying the first portion into the one or more blocks in the persistent memory ahead of a second portion of the virtual memory; and after copying the first portion: applying the write request to the first portion; and copying the second portion into the one or more blocks in the persistent memory.

Abstract: Methods, systems, and computer program products for high-performance cluster computing. Multiple components are operatively interconnected to carry out operations for high-performance RDMA I/O transfers over an RDMA NIC. A virtual machine of a virtualization environment initiates a first I/O call to an HCI storage pool controller using RDMA. Responsive to the first I/O call, a second I/O call is initiated from the HCI storage pool controller to a storage device of an HCI storage pool. The first I/O call to the HCI storage pool controller is implemented through a first virtual function of an RDMA NIC that is exposed in the user space of the virtualization environment. Prior to the first RDMA I/O call, a contiguous unit of memory to use in an RDMA I/O transfer is registered with the RDMA NIC. The contiguous unit of memory comprises memory that is registered using non-RDMA paths such as TCP or iSCSI.

Abstract: Various embodiments disclosed herein are related to a non-transitory computer readable storage medium. In some embodiments, the medium includes instructions stored thereon that, when executed by a processor, cause the processor to send an indication of a first storage location to a destination host. In some embodiments, the first storage location includes content that is swapped out from a memory location in a source host. In some embodiments, the indication includes one or more of a logical address and a first physical address. In some embodiments, the medium includes instructions stored thereon that, when executed by a processor, cause the processor to map the logical address of the first storage location to a second physical address of a second storage location. In some embodiments, the destination host accesses the content of the first storage location.

Abstract: Methods, systems, and computer program products for high-performance cluster computing. Multiple components are operatively interconnected to carry out operations for high-performance RDMA I/O transfers over an RDMA NIC. A virtual machine of a virtualization environment initiates a first I/O call to an HCI storage pool controller using RDMA. Responsive to the first I/O call, a second I/O call is initiated from the HCI storage pool controller to a storage device of an HCI storage pool. The first I/O call to the HCI storage pool controller is implemented through a first virtual function of an RDMA NIC that is exposed in the user space of the virtualization environment. Prior to the first RDMA I/O call, a contiguous unit of memory to use in an RDMA I/O transfer is registered with the RDMA NIC. The contiguous unit of memory comprises memory that is registered using non-RDMA paths such as TCP or iSCSI.

Abstract: Methods, systems, and computer program products for high-performance cluster computing. Multiple components are operatively interconnected to carry out operations for high-performance RDMA I/O transfers over an RDMA NIC. A virtual machine of a virtualization environment initiates a first I/O call to an HCI storage pool controller using RDMA. Responsive to the first I/O call, a second I/O call is initiated from the HCI storage pool controller to a storage device of an HCI storage pool. The first I/O call to the HCI storage pool controller is implemented through a first virtual function of an RDMA NIC that is exposed in the user space of the virtualization environment. Prior to the first RDMA I/O call, a contiguous unit of memory to use in an RDMA I/O transfer is registered with the RDMA NIC. The contiguous unit of memory comprises memory that is registered using non-RDMA paths such as TCP or iSCSI.

Abstract: A system and method receiving a request to transfer first data from a first storage space to a second storage space, receiving a write request to write second data to a location during the transfer of the first data, determining from an access data structure that the location is not in use, writing the second data to the second storage space, and updating a location data structure indicating the location of the second data to be in the second storage space.

Abstract: Communication between VMs running in a clustered virtualization environment and a distributed storage system is managed. The distributed storage system consists of multiple storage/compute nodes. A hypervisor communicates with a controller in a local node on behalf of a VM. The controller will redirect a connection to a second controller on different node. The controller to which the connection is redirected is selected so as to distribute the load evenly across the cluster.

Abstract: Systems and methods are disclosed herein for writing data, including storing, by a storage client, a cache copy of the data in a memory of the storage client and receiving, by the storage client, acknowledgement indicating that a single replica of the data is made in a storage node remote to the storage client.

Abstract: A system and method include initiating a live migration of a virtual machine from a first host machine to a second host machine. A shared host physical storage includes first swapped out memory data associated with the first virtual machine from a first memory of the first host machine, and metadata including location information of the first swapped out memory data, and an identity of the associated first virtual machine. The system and method include copying memory data associated with the first virtual machine stored in the first memory to the second host machine. The system and method also include accessing, by a second hypervisor at the second host machine, the metadata stored in the shared host physical storage to determine location of the first swapped out memory data associated with the first virtual machine.

Abstract: Systems and methods are disclosed herein for writing data, including storing, by a storage client, a cache copy of the data in a memory of the storage client and receiving, by the storage client, acknowledgement indicating that a single replica of the data is made in a storage node remote to the storage client.

Abstract: A system and method include receiving, at a host device, a request from a virtual machine to communicate with an emulated device. The host device establishes a socket connection between an emulator and the emulated device and communicates input-output messages via the socket connection from the virtual machine to the emulated device where the input-output messages use a virtual function input/output (VFIO) message protocol.

Abstract: Systems and methods for migrating an original instance of a virtual machine (VM) to a new instance of the VM within a same host include generating, by a hypervisor of the host, memory mapping corresponding to a memory state of the original instance of the VM, sharing the memory mapping with the new instance of the VM, and migrating to the new instance of the VM based on the memory mapping.

Abstract: An exemplary embodiment disclosed herein is a method including a virtual storage controller receiving an input/output (I/O) command using a native drive communications protocol from a software application on a user virtual machine running on the host machine; determining whether a local service virtual machine is unresponsive to the virtual storage controller; translating the I/O command from the native drive communications protocol to an IP-based storage communications protocol to create a translated I/O command, in response to the local service virtual machine being unresponsive to the virtual storage controller; sending the translated I/O command to a remote service virtual machine, wherein the remote service virtual machine is configured to send the translated I/O command to a local storage and receive a response from the local storage; and receiving the response from the remote service virtual machine.

Abstract: A system and method receiving a request to transfer first data from a first storage space to a second storage space, receiving a write request to write second data to a location during the transfer of the first data, determining from an access data structure that the location is not in use, writing the second data to the second storage space, and updating a location data structure indicating the location of the second data to be in the second storage space.