Microsoft Technology Licensing, LLC

Marcus Felipe Fontoura

Ricardo Gouvea Bianchini

Ioannis Manousakis

Eli Cortez

Christian L Belady

Husam Atallah Alissa

Alok Gautam Kumbhare

Anand Bonde

Thomas Moscibroda

Winston Allen Saunders

Ricardo Gouvêa Bianchini

Nithish Mahalingam

Mark Edward Shaw

Sean Michael James

Mark Eugene Russinovich

Ajay Mani

Bharath Ramakrishnan

Yunqi Zhang

Inigo Goiri Presa

Nisarg Tarakkumar Sheth

Kyle Edward Woolcock

Brijesh Warrier

Reza Azimi

William Clausen

Inigo Goiri

Ana-Maria Constantin

Eric Clarence Peterson

David Allen Dion

James Ernest Johnson

Ashish Raniwala

Ke Wang

David Thomas Gauthier

Brian A Janous

Yimin Deng

Yunus Mohammed

Osvaldo Patricio Morales

Avnish Kumar Chhabra

Aditya Bhandari

Pritesh Patwa

Sameh Elnikety

Mark Alan Monroe

Gopal Kakivaya

Stephan W Gilges

Chandrasekhar Pasupuleti

Brian Jacob Corell

Milan Vukosavljevic

Lalu Vannankandy Kunnath

Ehsan Nasr Azadani

Ziv Rafalovich

Girish Bablani

Sean Patrick Abbott

Georgios Prekas

Rodrigo Lemos De Assis

Upshur Bagwell Quinby, V

Kapil Arya

Alper Gun

Stephan Wayne Gilges

A thermal management system for cooling electronic devices includes an immersion cooling system, a vapor buffer tank, and a liquid buffer tank. The immersion cooling system includes an immersion tank defining an immersion chamber, a working fluid in the immersion chamber, and a condenser. A liquid portion of the working fluid defines an immersion bath in the immersion chamber and a vapor portion defines a headspace above the immersion bath in the immersion chamber. The condenser condenses the vapor portion of the working fluid to the liquid portion of the working fluid. The vapor buffer tank is in fluid communication with the headspace, and a vapor valve selectively allows fluid communication between the vapor buffer tank and the headspace. The liquid buffer tank is in fluid communication with the immersion chamber, and a liquid valve selectively allows fluid communication between the liquid buffer tank and the immersion chamber.

Systems and methods for immersion-cooled datacenters

A power controller allocates input power to a datacenter between computing power for computing services, backup power, and overhead power for overhead systems. The power controller reallocates the overhead power and/or the backup power to the computing power. This may increase the overall utilization of the datacenter by allowing additional processing power of the servers to be used.

Allocating power between overhead, backup, and computing power services

Systems and methods for machine learning-based power capping and virtual machine placement in cloud platforms are disclosed. A method includes applying a machine learning model to predict whether a request for deployment of a virtual machine corresponds to deployment of a user-facing (UF) virtual machine or a non-user-facing (NUF) virtual machine. The method further includes sorting a list of candidate servers based on both a chassis score and a server score for each server to determine a ranked list of the candidate servers, where the server score depends at least on whether the request for the deployment of the virtual machine is determined to be a request for a deployment of a UF virtual machine or a request for a deployment of an NUF virtual machine. The method further includes deploying the virtual machine to a server with highest rank among the ranked list of the candidate servers.

Machine learning-based power capping and virtual machine placement in cloud platforms

The present disclosure describes a type of virtual machine, which the present disclosure may refer to as a harvest virtual machine, that may allow improved utilization of physical computing resources on a cloud-computing system. First, the harvest virtual machine may be evictable. In other words, higher priority virtual machines may preempt the harvest virtual machine's access to physical computing resources. Second, the harvest virtual machine may receive access to a dynamic amount of physical computing resources during the course of its operating life. Third, the harvest virtual machine may have a minimum size (in terms of an amount of physical computing resources) and may terminate whenever the harvest virtual machine has access to an amount of physical computing resources less than the minimum size.

Harvest virtual machine for utilizing cloud-computing resources

Techniques of managing oversubscription of network resources are disclosed herein. In one embodiment, a method includes receiving resource utilization data of a virtual machine hosted on a server in a computing system. The virtual machine is configured to perform a task. The method also includes determining whether a temporal pattern of the resource utilization data associated with the virtual machine indicates one or more cycles of resource utilization as a function of time and in response to determining that the temporal pattern associated with the virtual machine indicates one or more cycles of resource utilization as a function of time, causing the virtual machine to migrate to another server that is not oversubscribed by virtual machines in the computing system.

Resource oversubscription based on utilization patterns in computing systems

The present disclosure relates to systems, methods, and computer readable media for enabling server devices to utilize a higher percentage of power resources while maintaining sufficient availability of power resources of a datacenter or other collection of server devices. For example, systems disclosed herein determine and implement power shaving actions based on virtual machine metadata and in accordance with a power shaving policy to facilitate a significantly higher utilization of power resources on a datacenter during normal operation as well as within periods of limited power capacity on various server devices. Systems described herein provide more efficient utilization of power resources while maintaining service availability guarantees for a variety of virtual machines hosted by servers of the datacenter.

Managing power resources for pools of virtual machines

The discussion relates to disaggregated computing. One example can monitor multiple two-phase liquid immersion tanks. Individual two-phase liquid immersion tanks can contain multiple components of a single type of component type. The example can receive requests for virtual machines and allocate sets of components from individual two-phase liquid immersion tanks to work together to support the virtual machines requests.

Disaggregated computer systems

The present disclosure relates to systems, methods, and computer readable media that utilize a low-impact live-migration system to reduce unfavorable impacts caused as a result of live-migrating computing containers between physical server devices of a cloud computing system. For example, systems disclosed herein evaluates characteristics of computing containers on server devices to determine a predicted unfavorable impact of live-migrating the computing containers between the server devices. Based on the predicted impact, the systems disclosed herein can selectively identify which computing containers to live-migrate as well as carry out live-migration of the select computing containers in such a way the significantly reduces unfavorable impacts to a customer or client device associated with the computing containers.

Minimizing impact of migrating virtual services

A system receives a request to deploy a virtual machine (VM) on one of a plurality of nodes running a plurality of VMs in a cloud computing system. The system receives a predicted lifetime for the VM and an indication of an average lifetime of VMs running on each of the plurality of nodes. The system allocates the VM to a first node when a first policy of collocating VMs with similar lifetimes on a node is adopted and the predicted lifetime is within a predetermined range of the average lifetime of VMs running on the first node. The system allocates the VM to a second node when a second policy of collocating VMs with dissimilar lifetimes on a node is adopted and the predicted lifetime is not within the predetermined range of the average lifetime of VMs running on the second node.

Method for deploying virtual machines in cloud computing systems based on predicted lifetime

A system receives a request to deploy a virtual machine on a node from a plurality of nodes running a plurality of virtual machines in a cloud computing system. The system selects one of the plurality of nodes having a hard disk drive (HDD) input output operations per second (IOPS) value less than an observed HDD IOPS value for the plurality of nodes running the plurality of virtual machines. The system receives a predicted HDD IOPS value for the virtual machine and determines a new HDD IOPS value for the selected node based on the HDD IOPS value for the selected node and the predicted HDD IOPS value for the virtual machine. The system instantiates the virtual machine on the selected node when the new HDD IOPS value for the selected node is less than or equal to the observed HDD IOPS value for the plurality of nodes.

Resource management for virtual machines in cloud computing systems

Various embodiments, methods, and systems for implementing a predictive rightsizing system are provided. Predicted rightsized deployment configurations are generated for virtual machine 'VM' deployments having deployment configurations that are modified to predicted rightsized deployment configurations based on a prediction engine. In operation, a VM deployment, associated with a request to deploy one or more VMs on a node, is accessed at a predictive rightsizing controller. A predicted resource utilization for the VM deployment is generated at the prediction engine and accessed at the predictive rightsizing controller. The predicted resource utilization is generated based on a prediction engine that uses past behaviors and features associated with previous VM deployments. Based on the predicted resource utilization, a predicted rightsized deployment configuration is generated for the VM deployment. The predicted rightsized deployment configuration is communicated such that the request to deploy the VMs is executed based on the predicted rightsized deployment configuration.

Predictive rightsizing for virtual machines in cloud computing systems

Growing community of inventors

Clyde Hill, WA, United States of America

Marcus Felipe Fontoura