Microsoft Technology Licensing, LLC

Microsoft Corporation

Jacob Kappeler Oshins

Dustin Lyle Green

Stephane Plante

Doron J Holan

Andrew John Thornton

John Andrew Starks

Lars Reuther

Brandon Allsop

Andrei Warkentin

Eric P Traut

Hadden Mark Hoppert

Jeffrey Kinsey

Tony Dwayne Pierce

Christopher Leonard Huybregts

Hari Angepat

Michael L Neil

Rene Antonio Vega

Nathan T Lewis

Christopher Leland Eck

Vadim Makhervaks

Adrian J Oney

Michael H Tsang

Benjamin A Leis

Andrew Ernest Nicholas

Peter L Johnston

Neal Robert Christiansen

Andrew J Ritz

Pankaj Garg

John D Vert

Gerardo Diaz-Cuellar

Omar Cardona

Nicholas Stephen Judge

Parag Chakraborty

Adam B Glass

Michael Patrick Hills

Gaurav Sinha

Yi Yuan

Lawrence Ralph Cleeton

Phani Chiruvolu

Craig Daniel Wilhite

Rich Yampell

Vincent J Geglia Ii

Devdeep Singh

David Richards

Tejas Karandakar

Christopher L Eck

Nathan Lewis

Embodiments of the present disclosure include systems and methods for providing a scalable controller for managing data storages. A system includes a non-volatile memory controller comprising a set of data queues and a set of administrative queues. The system also includes a set of physical storages communicatively coupled to the non-volatile memory controller. A set of logical storages are created from the set of physical storages. A primary non-volatile memory controller is created from the non-volatile memory controller. The primary non-volatile memory controller comprising an administrative queue in the set of administrative queues, a first subset of the set of data queues, and a first subset of the set of logical storages. An extended non-volatile memory controller is created from the non-volatile memory controller. The extended non-volatile memory controller comprising a second subset of the set of data queues and a second subset of the set of logical storages.

Scalable controller for managing data storages

Examples described herein generally relate to hosting virtual memory backed kernel isolated containers. A server includes at least one physical processor and at least one physical computer memory addressable via physical memory addresses. The at least one physical computer memory stores executable code configured to provide at least one host including a kernel and at least one kernel isolated container within the at least one host. The host allocates virtual memory having virtual memory addresses to a respective container of the at least one kernel isolated container. The host pins a subset of the virtual memory addresses to a subset of the physical memory addresses. The host performs a direct memory access operation or device memory-mapped input-output operation of the respective container on the subset of the physical memory addresses. At least part of the physical computer memory that is not pinned is oversubscribed.

Pinned physical memory supporting direct memory access for virtual memory backed containers

A system for capturing hardware telemetry includes a hardware component encoded with hardware logic for emitting a telemetry stream into memory of a computing device. The system further includes a hardware component driver stored in the memory that is configured to parse the telemetry stream, populate telemetry structures defined within a telemetry event schema based on values parsed from the telemetry stream, and generate a telemetry record including the populated telemetry structures.

Telemetry generation for in-field hardware testing

Described is a technology by which a virtual hard disk is migrated from a source storage location to a target storage location without needing any shared physical storage, in which a machine may continue to use the virtual hard disk during migration. This facilitates use the virtual hard disk in conjunction with live-migrating a virtual machine. Virtual hard disk migration may occur fully before or after the virtual machine is migrated to the target host, or partially before and partially after virtual machine migration. Background copying, sending of write-through data, and/or servicing read requests may be used in the migration. Also described is throttling data writes and/or data communication to manage the migration of the virtual hard disk.

Background migration of virtual storage

Techniques for graphics processing unit (GPU) partitioning for virtualization are described herein. In one or more implementations, a GPU partitioning manager of a host device obtains a request for a virtual machine having GPU functionality. In particular, the request specifies the GPU functionality in terms of different GPU capabilities. These different capabilities correspond to segments of a GPU model that represents GPU functionality and is used to govern interactions between virtual machines and GPUs. The GPU partitioning manager determines whether GPUs of the host device are available to satisfy the request based on the specified capabilities. If so, the GPU partitioning manager allocates a portion of the determined available GPUs to the virtual machine to configure the virtual machine with a GPU partition having the functionality. The virtual machine configured with the GPU partition can then be exposed to provide GPU-processed data to a GPU partition requestor.

Graphics processing unit partitioning for virtualization

Techniques for effectuating a virtual NUMA architecture for virtual machines are disclosed herein. In an embodiment, a system determines the physical topology of a datacenter. When instantiating a virtual machine, the system determines a number of NUMA nodes for that VM based on the physical topology. The system then directs a second system to instantiate the virtual machine with the determined number of NUMA nodes.

Instantiating a virtual machine with a virtual non-uniform memory architecture and determining a highest detected NUMA ratio in a datacenter

A technique is described herein for updating the logic used by a hardware accelerator provided by a computing device. In one implementation, the technique provides a pass-through mode which allows a virtual machine (provided by the computing device) to directly interact with the hardware accelerator. Upon the commencement of an updating operation, the technique instructs an emulator to begin emulating the function(s) of the hardware accelerator and the resultant effects of these functions, without interaction with the actual hardware accelerator. When the updating operation finishes, the technique re-enables the pass-through mode. By virtue of the above-summarized manner of operation, the technique allows the computing device to perform the function(s) associated with the hardware accelerator while the hardware accelerator is being updated. In one case, the technique disables the pass-through mode by modifying address-mapping information used by the virtual machine to access system physical addresses.

Updating hardware with reduced virtual machine downtime

Techniques for effectuating a virtual Non-Uniform Memory Architecture (NUMA) for virtual machines and adjusting memory in virtual NUMA nodes are described. The virtual NUMA node topology may include a plurality of virtual NUMA nodes. The memory may be adjusted based on pressure in a specific virtual NUMA node of the plurality and an amount of guest memory assigned to the specific virtual NUMA node.

Adjusting guest memory allocation in virtual non-uniform memory architecture (NUMA) nodes of a virtual machine

A computing device has first and second virtual machines (VMs) and a resource assigned to the first VM. Each access request for the resource is forwarded thereto until the first VM is to be saved or migrated. Thereafter, each access request is forwarded to a holding queue. When the resource has acted upon all access requests forwarded thereto, the resource is reassigned to the second VM, and each access request at the holding queue is forwarded to the second VM and then the resource. Thus, all access requests for the resource are acted upon by the resource even after the resource is removed from the first VM and assigned to the second VM, and the save or migrate of the first VM can thereafter be completed.

Migrating a virtual machine that owns a resource such as a hardware device

Techniques for recovering virtual machine state and boot information used to boot an installed guest operating system on systems where the information has either been lost or is not present are described.

Intelligent boot device selection and recovery

A virtual machine storage service can be use a unique network identifier and a SR-IOV compliant device can be used to transport I/O between a virtual machine and the virtual machine storage service. The virtual machine storage service can be offloaded to a child partition or migrated to another physical machine along with the unique network identifier.

Virtual storage target offload techniques

Embodiments relate to off-loading aspects of storage virtualization to storage hardware and modifying software to take advantage of hardware virtualization features. A co-design of hardware and software allows a filesystem to provide files such that indirection overhead normally needed to access the content of files can be bypassed while still managing the files as filesystem objects. A storage device manages and exposes a virtual volume which is used to store the content of a file. Virtual volumes can be initialized or populated so that virtual blocks therein align with device storage blocks. A virtual volume can be initialized and populated by parsing a virtual disk file to access virtual disk metadata, which is then used to determine and set features of the virtual volume.

Storage virtualization offload

Growing community of inventors

Seattle, WA, United States of America

Jacob Kappeler Oshins