203 results about "Swarm computing" patented technology

A system and method for sharing network resources; the system comprising: (1) at least one network switch; (2) a plurality of computing devices, where the computing device comprises (i) at least one network connection; (ii) a plurality of processing nodes; and (iii) at least one storage device containing software for (a) initializing and maintaining a plurality of top-layer virtual local area networks (VLANs), (b) initializing and maintaining a plurality of client VLANs; and (c) using an empty VLAN as a virtual wire between the client VLAN and a shared network resource.

Runtime connection load balancing of work across connections to a clustered computing system involves the routing of requests for a service, based on the current operational performance of each of the instances that offer the service. A connection is selected from an identified connection pool, to connect to an instance that provides the service for routing a work request. The operational performance of the instances may be represented by performance information that characterizes the response time and/or the throughput of the service that is provided by a particular instance on a respective node of the system, and is relative to other instances that offer the same service.

Techniques for fast recovery and/or balancing of connections to a clustered computing system provide management of such connections by determining a number of connections to load balance across nodes and by triggering creation of such connections. In one aspect, a notification of an event regarding the clustered computing system is received by a connection pool manager, a pool of connections to the system is identified based on the notification, and one or more connections from the pool are processed in response to the event. According to an embodiment, the notification comprises the identities of the service, database, server instance and machine that correspond to the event.

Methods and apparatus for performing a data migration in a clustered computer system. In one aspect, the availability of the data being migrated is maintained during the migration. In another aspect, a data set is migrated by a transfer utility in one of the hosts in the cluster, and in response to a failure of the host, the migration is resumed without restarting from the beginning point of the data set. In a further embodiment, the hosts in the cluster have agents associated therewith to manage the data migration, and prior to taking at least one action associated with the data migration, the agents coordinate to ensure that the hosts in the cluster are prepared to have the at least one action taken.

A method, apparatus, system, article of manufacture, and data structure provide the ability to perform a sorted map-reduce job on a cluster. A cluster of two or more computers is defined by installing a map-reduce framework onto each computer and formatting the cluster by identifying the cluster computers, establishing communication between them, and enabling the cluster to function as a unit. Data is placed into the cluster where it is distributed so that each computer contains a portion of the data. A first map function is performed where each computer sorts their respective data and creates an abstraction that is a representation of the data. The abstractions are exchanged and merged to create complete abstraction. A second map function searches the complete abstraction to redistribute and exchange the data across the computers in the cluster. A reduce function is performed in parallel to produce a result.

A preferred embodiment of the present invention provides a way of gathering performance data during execution of an application executing on clustered machines. This data is then presented to the user in a way that makes it east to determine what variables and situations to change in order to improve performance. A described embodiment displays the color of displayed vertical bands in accordance with how much a particular call is contributing to the effect of communication time, latency, and bandwidth.

Improved techniques for controlled take over of services for clustered computing systems are disclosed. The improved techniques can be implemented to allow one sub-cluster of the clustered computing system to safely take over services of one or more other sub-clusters in the clustered computing system. Accordingly, if the clustered computing system is fragmented into two or more disjointed sub-clusters, one sub-cluster can safely take over services of the one or more other sub-clusters after the one or more other sub-clusters have been shutdown. As a result, the clustered computing system can continue to safely provide services even when the clustered computing system has been fragmented into two or more disjointed sub-clusters due to an operational failure.

The invention provides an anisotropic three-dimensional prestack time migration method which is applied to processing reflection seismic data in seismic exploration and relates to a prestack time migration method aiming at three-dimensional seismic data acquisition. The method takes into account the effect of the velocity anisotropy of an earth medium on the travel time and the amplitude value propagated by a seismic wave and can independently determine the migration velocity and the anisotropic parameters during the migration process, therefore, a migration image which is accurately homed and is preserved in amplitude is obtained. The method determines the three-dimensional time-varying migration aperture in an anisotropic medium according to an underground structure time-varying inclination angle and can simultaneously press the migration noise during the migration process. With the adoption of the method, the efficient and parallel calculation is realized through reasonable distribution of the seismic data among compute nodes of a cluster computer. The method has the key point that the travel time, the amplitude value and the imaging weight coefficients of the seismic wave and an incident angle in the anisotropic medium are solved with the application of a one-way wave operator with deep migration and the steady phase point principle; and has the important application value for oil-gas and mineral resource exploration.

The invention discloses a clustered image splitting method based on particle swarm optimization and spatial distance measurement. The clustered image splitting method mainly solves the problem that the existing clustered image splitting technology has local misdivision phenomena and multiple area miscellaneous points. The clustered image splitting method includes the steps: (1) inputting an original image, extracting pixel characteristics and conducting watershed splitting, (2) calculating an adjacent matrix according to a split area and generating clustered data, (3) using the clustered data to initialize a cluster at random, (4) calculating membership matrixes and adaptability values of the cluster, upgrading individuals and overall to be optimum, and evolving the cluster, (5) upgrading iterations, outputting the best membership matrix if the preset maximum iterations are achieved, and continuously executing the step (4) if the preset iterations are not achieved, and (6) according to the best membership matrix, marking according to the maximum probability principle to obtain the splitting result. Compared with the prior art, the clustered image splitting method has the advantages of being good in area coincidence, high in splitting accuracy and capable of being used for object identification on an SAR image.

The invention discloses a parameter communication optimization method for distributed machine learning. According to the method, the fault-tolerant characteristic of the machine learning iteration-convergence algorithm is expanded; a dynamic finite fault tolerance characteristic is provided, a distributed machine learning parameter communication optimization strategy is realized based on the dynamic finite fault tolerance, the performance of each computing node is fully utilized by dynamically adjusting the synchronization strategy of each computing node and a parameter server in combination with a performance detection model, and the accuracy of the machine learning model is ensured; sufficient computing resources are guaranteed, and the training process of the model is not affected by dynamic changes of the distributed computing resources; a training algorithm and system hardware resources are decoupled, the process that developers manually allocate computing resources and adjust andoptimize data communication according to experience is liberated, and the expansibility and high execution efficiency of a program in various cluster environments are effectively improved. The methodcan be applied to the fields of optimization of distributed machine learning parameter communication, optimization of cluster computing performance and the like.

The present invention discloses a multi-component self-organizing software-connected cluster computer intelligent resource management method. The method gradually increases the component resources along with the increase of user number and gradually reduces the component resources along with the decrease of user number according to the component resources provided by the user to the dynamic demand of the resources, thus achieving the automatic start-up and the automatic shutdown of the components. Through monitoring the use of the component resources, the method achieves the dynamic scheduling of the component resources in the component cluster, carries out the fault management and safety management of the components, and monitors and maintains the reliable operation of the component resources. The present invention also carries out the partitioning treatment of system storage resources and satisfies the temporary and permanent storage requirements of different user data. The present invention can increase the utilization rate of the component resources and optimizes the resource configuration of the cluster computer, thus further decreasing the investment cost and the operating maintenance and management cost, and increasing the system reliability.

The embodiment of the invention discloses a task processing method and a device, a server, a client, a system and a storage medium, and the task processing method comprises the steps: receiving a tasksubmission request sent by the client, and the task submission request comprises the configuration information of a task; generating a task submission script containing configuration information; andsubmitting the task submission script to a computing cluster, wherein the computing cluster is used for running the task according to the configuration information in the task submission script. According to the method of the invention, task submission is divided into a client and a server; a user only needs to configure configuration information during task running at a client; a user does not need to deeply learn meanings of various parameters of the computing cluster; therefore, on one hand, the learning cost of the user and the threshold that the user uses the computing cluster are reduced, and on the other hand, the user realizes safe isolation between the computing cluster and the user side through indirect reaction between the computing cluster and the server, and a guarantee is provided for the security of the computing cluster.

A node manager (300) resides on a node (104) in a cluster computing system (100) and transfers information and events being communicated across the node (104) to a broker (102) coupled to the node manager (300). The broker (102) transmits information to client devices (106) who subscribe to particular events. The client devices (106) publish their own messages back to the broker (102) or to a second broker (1002). Other client device (1102) can then subscribe to receive the messages from either broker (102 or 1002). The other client devices (1102) can then publish their own messages back to the brokers for subscription by further client devices (902).

A scalable, malware detection system features at least one sensor and a cluster including at least one computing node. The computing node includes an analysis coordination system and an object analysis system. The analysis coordination system, when activated as a broker computing node, (i) receives metadata from a sensor, (ii) analyzes the metadata, and (iii) places at least a portion of the metadata into a data store for subsequent use in retrieval of the suspicious object by the object analysis system from the sensor. The object analysis system is configured to (i) retrieve the portion of the metadata, which includes at least a sensor identifier, from the data store, (ii) retrieve the suspicious object from the sensor using at least part of the portion of the metadata retrieved from the data store, and (iii) analyze the suspicious object for malware.

The invention provides a guided mirror image making method and device, electronic equipment and a storage medium, belongs to the technical field of computer networks, and solves the problems that if many cluster computing nodes exist, users are configured one by one, time is greatly wasted, virtual machine creation failure is caused easily due to missed configuration, If a user does not know whether a used operating system supports UEFI starting or not, mirror image manufacturing fails, repeated manufacturing exists, and a large amount of time of the user is wasted. The method comprises the steps of querying whether a computing node of a cloud platform has an OVMF installation file or not; selecting an operating system supporting UEFI startup; and respectively manufacturing a UEFI boot mirror image and a BIOS boot mirror image. According to the method, the double-boot mirror image is manufactured on the cloud platform of the OpenStack, whether the OVMF software is installed in the computing node of the server or not is checked in a one-key mode, whether the version of the operating system supports UEFI starting or not is intelligently verified, the system can be automatically switched to the needed boot mode to create the virtual machine, the mirror image manufacturing efficiency is improved, and a large amount of time is saved.

Disclosed herein are system, apparatus, article of manufacture, method, and/or computer program product embodiments for monitoring long-standing computing resources. An apparatus may operate by receiving a cloud monitoring notification, where the cloud monitoring notification may indicate an occurrence of a monitored condition. The apparatus may then operate by scanning a cluster computing system for resource having a client assigned resource identifier and a computing resource attribute based on a resource identifier scan parameter and a resource attribute scan parameter. The apparatus may further operate by generating a resource notification request based on the scanning of the cluster computing system and transmitting the resource notification request to a communications system to notify a user that the resource has a computing resource attribute that match the resource attribute scan parameter.