公开(公告)号：US20250004863A1

公开(公告)日：2025-01-02

申请号：US18346069

申请日：2023-06-30

Applicant: Amazon Technologies, Inc.

Inventor： Christopher Denis Scalabrini ,  Adam Julio Villalobos

IPC: G06F9/52 ,  G06F9/50

Abstract: Automated scaling-related operations may be performed dynamically during execution of a spatial simulation. A spatial partition may be locally reassigned, based on application workload information, from a first application to a second application on the same worker. A quantity of applications on a worker may also be changed during execution of a spatial simulation. A parent spatial partition may be split into child spatial partitions, and child partitions may also be merged back into a common parent partition. Indications of partition splits and merges on each of a plurality of workers may be reported to the plurality of workers. A spatial partition may also be remotely reassigned from a first worker to a second worker, such as based on worker-level resource consumption information and partition information. A quantity of workers that are used to implement a spatial simulation may also be changed during execution of the spatial simulation.

公开(公告)号：US11416553B2

公开(公告)日：2022-08-16

申请号：US16368677

申请日：2019-03-28

Applicant: Amazon Technologies, Inc.

Inventor： Eric Jason Edelman ,  Christopher Denis Scalabrini ,  Adam Julio Villalobos

IPC: G06F16/00 ,  G06F16/901 ,  G06F16/27 ,  G06F16/44 ,  G06F16/84 ,  G06F16/909 ,  G06F16/906 ,  G06F16/28 ,  G06F9/50 ,  G06F16/22

Abstract: Technology is described for using spatial analysis of data objects associated with a multi-dimensional virtual environment to organize computational units in a distributed computing system. The data objects may be grouped together using spatial location information and collections of data objects may be treated as processing partitions (i.e., sub-divisions of the entire group of data objects). A mapping may be created between the data objects and processing partitions for processing applications which process data objects assigned to a processing partition. The processing partitions may be defined, in part, according to how the data objects are grouped together spatially. The processing partitions may be load-balanced across hardware hosts in the distributed computing system.

公开(公告)号：US10826971B1

公开(公告)日：2020-11-03

申请号：US14582038

申请日：2014-12-23

Applicant: Amazon Technologies, Inc.

Inventor： Guido Beat Zgraggen ,  Christopher Ryan Mhley ,  Darren James Alton ,  Adam Julio Villalobos ,  Eric Benjamin Merritt ,  Xiao Zeng

IPC: G06F15/16 ,  H04L29/08

Abstract: Techniques for distributed computing system node management are described herein. In some cases, internal compute nodes (i.e., compute nodes that are allocated to the distributed system) may be mutually trusted such that they may freely establish communications with one another. By contrast, external compute nodes (i.e., compute nodes that aren't allocated to the distributed computing system) may be untrusted such that their access to the distributed system may be regulated. In some cases, one or more of the compute nodes within the distributed computing system may maintain respective collections of system view information. Each respective collection of system view information may include, for example, information associated with the corresponding compute node's view of the distributed computing system based on information that is available to the corresponding compute node.

公开(公告)号：US10133496B1

公开(公告)日：2018-11-20

申请号：US15182922

申请日：2016-06-15

Applicant: Amazon Technologies, Inc.

Inventor： Juan Pablo Ferreyra ,  Brian David Fisher ,  Adam Julio Villalobos ,  Yu Ping Hu

IPC: G06F12/00 ,  G06F3/06 ,  G06T11/60 ,  G06T7/73 ,  G06T19/00

Abstract: The disclosed techniques may employ components referred to herein as atoms for computing and maintaining of states. Unlike traditional actors, atoms may be capable of binding to other atoms to form a bound combination of atoms, referred to herein as a molecule. In some examples, while bound to other atoms, an atom may operate in a manner that is different from traditional actors. For example, in some cases, atoms that are bound to one another may be prohibited from concurrently performing different operations on their own separate states. Additionally, bound atoms may be operable to collectively (e.g., synchronously) perform shared operations on their associated states. Furthermore, a shared operation performed on the states of bound atoms may be performed atomically. Also, in some examples, bound atoms may be capable of communicating synchronously with one another and of synchronously accessing each other's states.

公开(公告)号：US20250001292A1

公开(公告)日：2025-01-02

申请号：US18346021

申请日：2023-06-30

Applicant: Amazon Technologies, Inc.

Inventor： Christopher Denis Scalabrini ,  Adam Julio Villalobos

IPC: A63F13/355 ,  A63F13/358

Abstract: Automated scaling-related operations may be performed dynamically during execution of a spatial simulation. A spatial partition may be locally reassigned, based on application workload information, from a first application to a second application on the same worker. A quantity of applications on a worker may also be changed during execution of a spatial simulation. A parent spatial partition may be split into child spatial partitions, and child partitions may also be merged back into a common parent partition. Indications of partition splits and merges on each of a plurality of workers may be reported to the plurality of workers. A spatial partition may also be remotely reassigned from a first worker to a second worker, such as based on worker-level resource consumption information and partition information. A quantity of workers that are used to implement a spatial simulation may also be changed during execution of the spatial simulation.

公开(公告)号：US20200310884A1

公开(公告)日：2020-10-01

申请号：US16368738

申请日：2019-03-28

Applicant: Amazon Technologies, Inc.

Inventor： Adam Julio Villalobos ,  Christopher Denis Scalabrini ,  Eric Jason Edelman

IPC: G06F9/50 ,  G06F9/38 ,  G06F16/9035 ,  G06F16/909 ,  G06F16/23

Abstract: Technology is described for computationally intensive distributed computing systems and methods. A method for using object types to distribute processing may include determining object types for data objects. A mapping between the data objects and processing partitions may be determined using the object types. A processing partition may be associated with a processing application that processes data objects of a given object type. A hardware host in a distributed computing system may identify an application library having a plurality of processing applications that process a plurality of data objects associated with a plurality of object types. The hardware host may use the processing partition assigned to the hardware host to identify the processing application for the object type associated with the processing partition from the application library. The hardware host may launch an instance of the processing application to enable the hardware host to process the plurality of data objects.

公开(公告)号：US10742718B1

公开(公告)日：2020-08-11

申请号：US14582006

申请日：2014-12-23

Applicant: Amazon Technologies, Inc.

Inventor： Guido Beat Zgraggen ,  Christopher Ryan Mhley ,  Darren James Alton ,  Adam Julio Villalobos ,  Eric Benjamin Merritt ,  Xiao Zeng

IPC: G06F15/16 ,  H04L29/08

Abstract: Techniques for distributed computing system node management are described herein. In some cases, internal compute nodes (i.e., compute nodes that are allocated to the distributed system) may be mutually trusted such that they may freely establish communications with one another. By contrast, external compute nodes (i.e., compute nodes that aren't allocated to the distributed computing system) may be untrusted such that their access to the distributed system may be regulated. In some cases, one or more of the compute nodes within the distributed computing system may maintain respective collections of system view information. Each respective collection of system view information may include, for example, information associated with the corresponding compute node's view of the distributed computing system based on information that is available to the corresponding compute node.

公开(公告)号：US10656967B1

公开(公告)日：2020-05-19

申请号：US15935451

申请日：2018-03-26

Applicant: Amazon Technologies, Inc.

Inventor： Cheyne William Anderson ,  Adam Julio Villalobos ,  Qiang Fan

IPC: G06F9/46 ,  G06F9/50 ,  G06F9/48

Abstract: Techniques for actor and thread message dispatching are described herein. In some examples, a computing system may include multiple actors and multiple processing threads. Upon receiving an incoming message associated with one of the actors, a dispatching component may select one of the multiple processing threads within the computing system to which to dispatch the message for processing. If the associated actor is already assigned to one of the processing threads, then that processing thread may be selected for processing of the message. By contrast, if the associated actor is not already assigned to one of the processing threads, then a processing thread may be selected based on various factors, such as a set of priorities associated with the processing threads.

公开(公告)号：US10243870B1

公开(公告)日：2019-03-26

申请号：US14582051

申请日：2014-12-23

Applicant: Amazon Technologies, Inc.

Inventor： Darren James Alton ,  Xiao Zeng ,  Adam Julio Villalobos ,  Eric Benjamin Merritt ,  Guido Beat Zgraggen ,  Christopher Ryan Mhley

IPC: H04L12/911

Abstract: Techniques for distributed computing system node management are described herein. In some cases, internal compute nodes (i.e., compute nodes that are allocated to the distributed system) may be mutually trusted such that they may freely establish communications with one another. By contrast, external compute nodes (i.e., compute nodes that aren't allocated to the distributed computing system) may be untrusted such that their access to the distributed system may be regulated. In some cases, one or more of the compute nodes within the distributed computing system may maintain respective collections of system view information. Each respective collection of system view information may include, for example, information associated with the corresponding compute node's view of the distributed computing system based on information that is available to the corresponding compute node.

公开(公告)号：US09959143B1

公开(公告)日：2018-05-01

申请号：US14804521

申请日：2015-07-21

Applicant: Amazon Technologies, Inc.

Inventor： Cheyne William Anderson ,  Adam Julio Villalobos ,  Qiang Fan

IPC: G06F9/46 ,  G06F9/50 ,  G06F9/48

CPC classification number: G06F9/544 ,  G06F9/4881

Abstract: Techniques for actor and thread message dispatching are described herein. In some examples, a computing system may include multiple actors and multiple processing threads. Upon receiving an incoming message associated with one of the actors, a dispatching component may select one of the multiple processing threads within the computing system to which to dispatch the message for processing. If the associated actor is already assigned to one of the processing threads, then that processing thread may be selected for processing of the message. By contrast, if the associated actor is not already assigned to one of the processing threads, then a processing thread may be selected based on various factors, such as a set of priorities associated with the processing threads.