ProActive
search
Trainings
CLOUD and HPC with OW2 Open Source ProActive Parallel Suite:
Location: Location: Paris (Luxembourg)
Schedule: February, 2013
News
PACAGrid : 1368 Cores Production Cloud Portal powered by ProActive
Careers
2 Engineer positions available !
Dec. 12, 2012
ProActive V5.3.2 is out!

Studio 1.2

The goal of ProActive Matlab Connector is to equip Matlab with a generic interface to Grid and Cloud computing. It is composed of a Matlab toolbox. It allows the deployment of Matlab instances on several nodes of various infrastructures, and to use these instances like computing engines. The submission of Matlab tasks to be distributed over nodes is achieved seamlessly and directly from the usual Matlab interactive environments.

This package features :

  • Run functions on remote computers, Grids, and Clouds.

  • Do not block the local session while remote results are being produced.

  • Seamlessly retrieve results when you need them, just as if the functions were run locally.

  • Provide you with detailed remote log/output information, altogether with errors if any occurred.

  • Disconnected mode, where the local session doesn't need to remain open while the job is processing.

  • Automatic source transfer, data file transfer, transfer of local workspace, and other configurable options.

Matlab Connector 1.1

The goal of ProActive Scilab Connector is to equip Scilab with a generic interface to Grid and Cloud computing. It is composed of a Scilab toolbox. It allows the deployment of Scilab instances on several nodes of various infrastructures, and to use these instances like computing engines. The submission of Scilab tasks to be distributed over nodes is achieved seamlessly and directly from the usual Scilab interactive environments.

This package features :

  • Run functions on remote computers, Grids, and Clouds.

  • Do not block the local session while remote results are being produced.

  • Seamlessly retrieve results when you need them, just as if the functions were run locally.

  • Provide you with detailed remote log/output information, altogether with errors if any occurred.

  • Disconnected mode, where the local session doesn't need to remain open while the job is processing.

  • Automatic source transfer, data file transfer, transfer of local workspace, and other configurable options.

Scilab Connector 1.1

The ProActive Scheduler allows easy distribution of all your applications (Native, Java, Scripts, Matlab, Scilab, GPU, etc.), together withDynamic Workflows. It provides automatic input and output file transfer, fault tolerance, resource selection and configuration through scripting (Javascript, Ruby, Python), multi-core and topology control.

Scheduler 3.3

ProActive Java Active Objects (Programming) is an open source Java library aiming to simplify the programming of multithreaded, parallel, and distributedapplications for Clouds, Grids, multi-cores, clusters, and data-centers. 

Java Active Objects features a concurrent and parallel programming model, offers distributed and asynchronous communications, mobility and a deployment framework.

Java Active Objects 5.3

The Scheduling and Resource Management Rest APIs provide a language-independent access to the Scheduler.

Use this interface to seamlessly integrate orchestration and scheduling in you own portal.

Main Features:

  • Jason serialization

  • War archive for direct integration into your Web Application

 

Rest API 1.3

The Scheduler Web portal provides a platform-independent access to the Scheduler.

Main Features:

  • Full visualization of Job and Task details

  • Full access to output and logs in streaming mode

  • GUI and Activation for Data Servers (File Transfer)

  • Remote Visualization of Graphical Tasks

  • Graphical monitoring of Workflow execution directly from the portal.

 

Scheduler Web Portal 1.3

The Cloud & Grid IaaS Web portal provides a platform-independent access to the Resource Manager.

Main Features:

  • Full visualization of resources, physical and virtual machines
  • Full control on resources, deploy or remove resources on desktop PCs, clusters, clouds...
  • Graphical monitoring of resource usage
Cloud & Grid IaaS Web Portal 1.3

The ProActive Cloud & Grid IaaS Server is the software for coupling distributed resources in order to solve large-scale problems. The distributed resources are represented by Java virtual machines launched on desktop computers, clusters or clouds.

The Cloud & Grid IaaS Server provides a single point of access to all resources enabling an effective way of selecting them for computations with different criteria.

 

Resource Manager 3.3

The ProActive Linux Agent allows to automatically start and monitor resources on Linux machines.

Main Features:

  • Automatic start, stop according to a weekly schedule.

  • Automatic monitoring and restart of forked processes.

  • Limit available system resources (cpu, I/O, memory).

  • Integrated with ProActive Cloud & Grid IaaS.

  • Same XML configuration file than the ProActive Windows Agent.

  • Can be started as a system daemon or as a regular executable without any special permissions.

     

 

Linux Agent 1.0

ProActive Windows Agent is a Windows Service: a long-running executable designed not to require user intervention. The agent is able to create a ProActive computational resource on the local machine. This resource will be provided to ProActive applications (such as the Cloud & Grid IaaS Server) according to the user defined planning.

Main features:

  • User defined weekly planning.

  • Local Registration of a ProActive node.

  • Cloud & Grid IaaS Server registration of a ProActive node.

  • Automatic restart of the spawned java process in case of failures.

  • RAM limitation of the spawned processes.

  • CPU limitation (throttling) of the spawned processes.

 

 

Windows Agent 2.4

Edited by:


With contributions from:
inriaunice

Get Professional Support

ProActive at a glance ... move the cursor over text and boxes for interactive information

Workflow & Parallelization
HPC Workflows, Application Parallelization
(End-User and Application Interfaces)


Studio

 
Web Portals

Java Active Objects
IC2D
 
APIs: Java, CLI, REST
Matlab & Scilab
MapReduce

Orchestration & Scheduling
Multi-platform Job & Workflow Scheduler
(Servers and Tools for Admin)

Orchestration & Scheduling
Server
 

REST Server
 

Web Servers
 
Scheduler Worker

Clouds & Grids IaaS
Heterogeneous Resource Manager with Physical and Virtual Machines
(Servers and Tools for Admin)

Clouds & Grids IaaS
Server
 

Linux Agent
 

Windows Agent
 
Virtual Machine Management
 
     
   

Workflows & Parallelization : ProActive Studio

ProActive Studio is a graphical environment which integrates :

Workflow Design

Orchestration & Scheduling GUI

Cloud & Grid IaaS GUI
A graphical environment for constructing, validating, customizing and documenting workflows.
A workflow is a combination of tasks, their dependencies and control statements define parallel computations.
Instead of manually creating a workflow xml file you can easily design it with a tool palette provided by the studio.		 The graphical interface of
ProActive Orchestration & Scheduling Server.
The graphical interface allows to easily submit new jobs, monitor activity and view jobs results using the appropriate application.		 The graphical interface of
ProActive Cloud & Grid IaaS Server.
The graphical interface allows to easily acquire new computing resources or release some, monitor the activity of resources and define resources’ acquisition policies.

Workflow & Parallelization : MapReduce

ProActive MapReduce is a framework for execution of MapReduce jobs using the infrastructure provided by ProActive Orchestration & Scheduling and ProActive Cloud & Grid IaaS Servers. The ProActive Servers provide fault tolerance, easy deployment and advanced scheduling capabilities not found in Hadoop. There is no distributed file system in ProActive though, so the DataSpaces mechanism is used for accessing input and writing output data.

ProActive MapReduce API is Hadoop-like. The user can define the ProActive MapReduce workflow as in Hadoop, but, internally, the ProActive MapReduce builds a ProActive job and submits it to the ProActive Orchestration & Scheduling Server. Adapting existing Hadoop jobs to execute them using ProActive MapReduce framework is easy. No changes are required to the Mapper and Reducer classes; and only minimal amount of additional configuration is needed.

Workflow & Parallelization : Matlab/Scilab

The goal of this package is to equip Matlab and Scilab with a generic interface to Grid and Cloud computing. This extension allows the deployment of Matlab and Scilab instances on several nodes of various infrastructures, and to use these instances like computing engines. The submission of both Matlab and Scilab tasks to be distributed over nodes is achieved seamlessly and directly from the usual Matlab and Scilab interactive environments.

This package features for both Matlab and Scilab:

  • Run functions on remote computers, Grids, and Clouds.

  • Do not block the local session while remote results are being produced.

  • Seamlessly retrieve results when you need them, just as if the functions were run locally.

  • Provide you with detailed remote log/output information, altogether with errors if any occurred.

  • Disconnected mode, where the local session doesn't need to remain open while the job is processing.

  • Automatic source transfer, data file transfer, transfer of local workspace, and other configurable options.

Workflow & Parallelization : Java Active Objects APIs
Master-Worker : Task-Based API for solving Master-Worker type computations (one Master, many workers).
Monte-Carlo : a Task-based API similar to Master-Worker, specialized for monte-carlo simulations Legacy Code Wrapping : Efficient methods to wrap, couple, and gridify legacy code with ProActive, including MPI or other native codes
 SPMD: SPMD technique is ideal for solving tightly-coupled parallel problems requiring intensive communications (CFD, Maxwell,...). This is an alternative to MPI in Java
 File Transfer: Files can be transmitted from or to a remote node at different stage of the application lifecycle

 Web Services: ProActive Programming Active Objects and GCM components can be exposed as web services without any additional development
Data Spaces : a virtual file system layer that allow files to be accessed among all Active Objects, accross networkss and computers
 Security: Communications can be encrypted using a private/public key mechanism to ensure fully secure data transmissions.
GCM: Grid Component Model API is based on and extends the Fractal model to enable component programming in ProActive. Fault Tolerance: Application can be configured to be self-healing and restart automatically from latest valid checkpoint.
Asynchrony : Calls on Active Objects are asynchronous: an answer, a future, is provided without waiting for the end of the result processing. Mobility: Active Objects have the ability to migrate between nodes while the application is running.
 Remote Accessibility: Methods defined by active objects can be called from any point of the application.

Workflow & Parallelization : IC2D GUI for Active Objects

Quickly find out the points of improvement in your applications and further accelerate them with ProActive Graphical tool, IC2D: monitor and visualize your system activity! It is a generic tool suite that can be easily integrated in your environment and extended through Eclipse RCP plug-ins: TimIt, CharIt, Extended Debugger.

Workflow & Parallelization : APIs (Java, REST, CLI)

The Orchestration & Scheduling and Cloud & Grid IaaS Servers APIs provide different language accesses to a ProActive Parallel Suite Platform.

Java REST CLI
Use this interface for a direct Java access to Orchestration & Scheduling and Cloud & Grid IaaS Servers. Use this interface for a language independant access of both Orchestration & Scheduling and Cloud & Grid IaaS in your applications and in your own custom CLOUD portal. The Command-Line Interfaces to Orchestration & Scheduling and Cloud & Grid IaaS Servers provide both user-level and admin-level accesses.

Main Features:

  • Comprehensive Java API (user and admin-level)

  • Event notification mechanism for Job completion, Resource acquisition, etc.

Main Features:

  • Jason serialization

  • War archive for direct integration into your Web Application

Main Features:

  • Comprehensive console environment including history, completion and detailed help

  • Scripting capabilities (execution of command scripts)

Workflow & Parallelization : Web Portals

The Orchestration & Scheduling and Cloud & Grid IaaS Web Portals provide a zero-configuration, real-time, graphical Web access to a ProActive Parallel Suite Platform.

Orchestration & Scheduling Web Portal Cloud & Grid IaaS Web Portal

Workflow & Parallelization

Java Parallel Toolkit

ProActive Programming is an open source innovative Java library that provides a powerful framework to easily develop parallel and distributed applications.

ProActive Programming hides complex aspects of parallel programming and enables the developers to focus on the business code as it takes care about the rest!

Accelerate & Scale your most demanding applications with Multi-Core and Distributed Parallel Computing:

  • Portability, Fault-tolerance, File transfer, SOA, Security.
  • Master Worker, Monte Carlo, Legacy Wrapping, SPMD, File Split & Merge, Task Flow.
  • Asynchrony, groups and futures.
  • Monitoring & Optimizing GUI.

 

 See PDF Product Sheet
 See PDF Data Sheet
 Download

Workflow & Parallelization : IC2D GUI for Active Objects

Quickly find out the points of improvement in your applications and further accelerate them with ProActive Graphical tool, IC2D: monitor and visualize your system activity! It is a generic tool suite that can be easily integrated in your environment and extended through Eclipse RCP plug-ins: TimIt, CharIt, Extended Debugger.

Orchestration & Scheduling Server

Multi-platform Job & Workflow Scheduler

ProActive Orchestration & Scheduling Server is an open source multi-platform Job Scheduler managing the distribution of taskflows and applications execution over the available computing resources.

ProActive Orchestration & Scheduling ensures more work done with fewer resources: maximum utilization and optimal allocation of existing IT infrastructure, reducing administration costs and future hardware expenditures.

Schedule and manage parallel jobs on all your company’s resources and on external clouds:

  • Fault-tolerance, Security, File Transfer
  • Taskflows in Java or XML with temporal and data dependencies
  • Java or native applications, scripts (Ruby, Python, JavaScript), Matlab, Scilab, MPI, GPU, ProActive Programming
See PDF Product Sheet
See PDF Data Sheet
Download

Orchestration & Scheduling : Web Servers

ProActive Web Servers run next to ProActive Orchestration & Scheduling and Cloud & Grid IaaS Servers and allow web access to them via Web Browsers in Portal mode.

Orchestration & Scheduling : REST Server

ProActive REST Server> runs next to ProActive Orchestration & Scheduling and Cloud & Grid IaaS Servers and allows a language independant access via the RESTs API.

Orchestration & Scheduling : Scheduler Worker

The Scheduler Worker package allows to deploy scheduler worker code for easily including Cloud & Grid IaaS nodes. Thus, allowing to use the Nodes from both Cloud & Grid IaaS and Orchestration & Scheduling. Note that Linux and Windows Agents are also available (see below in the Figure).

Cloud & Grid IaaS Server

Heterogeneous Resource Manager with Physical and Virtual Machines

ProActive Cloud & Grid IaaS provides an open source intelligent and adaptive application deployment engine to virtualize the hardware resources, monitor and control all computing resources. Resources management gets easier and highly configurable.

ProActive Cloud & Grid IaaS leverages organizations existing infrastructure, from Desktops, dedicated clusters to heterogeneous distributed resources and Private Clouds.

ProActive Cloud & Grid IaaS allows you to:

  • Build your Enterprise Cloud with Virtualization (VMware, KVM, Xen)
  • Connect with external Public Clouds
  • Deploy applications with various protocols (ssh, rsh, sshGSI), on most schedulers (PBS, LSF, SGE, Prun, EGEE gLite, IBM LL)
  • Use Windows and Linux Agents
  • Accounting of resource usage
  • Cloud Burst (EC2, Azure, private Data Centers)
See PDF Product Sheet
See PDF Data Sheet
Download

 

Clouds & Grids : ProActive Windows Agent

The windows agent allows computers to be efficiently used both as desktop machines and computing resources. It's a small program running in the background which makes the computer available as a resource to the ProActive Cloud & Grid IaaS Server whenever it's not used (e.g. no foreground operation is active and the screen saver is on). It can as well be configured according to any timing/schedule policy.

Clouds & Grids : ProActive Linux Agent

The ProActive Linux Agent allows to automatically start and monitor resources on Linux machines.

Main Features:

  • Automatic start, stop according to a weekly schedule.

  • Automatic monitoring and restart of forked processes.

  • Limit available system resources (cpu, I/O, memory).

  • Integrated with ProActive Cloud & Grid IaaS Server.

  • Same XML configuration file than the ProActive Windows Agent.

  • Can be started as a system daemon or as a regular executable without any special permissions.

     

Clouds & Grids :
ProActive Virtual Machine Management

ProActive Virtual Machine Management is a virtualization API, hypervisor agnostic, which provides users with the capability to automate virtual machine deployment. It leverages hypervisors's native APIs and brings them to Java.

Main features:

  • Start/Stop/Clone/Destroy Virtual Machines.

  • PushData/GetData (update VM's environment with Key/Value pairs).

  • Supports VMware products: Workstation >= 6.5, ESX >= 3, Server >= 1.5

  • Supports Sun xVM VirtualBox >= 2.2

  • Supports XenServer >= 4

  • Supports Microsoft Server 2008 R2 Hyper-V

  • Implements a Libvirt bridge.

 
© 2013 Inria Sophia Antipolis Login