SixFaceCloud Installation

221 views

Published on

SixFaceCloud is a platform for solving financial analytics and quantitative analysis tasks using distributed computing.
SixFaceCloud installation guide.
http://www.sixfacecloud.com/

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
221
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

SixFaceCloud Installation

  1. 1. System Installation and ConfigurationIntroductionSixFaceCloud is a platform for solving financial analytics and quantitative analysis tasks using distributedcomputing. The solution consists of two main parts: Execution Engine and development tools. Facilitateapplication development using advanced tools; run your financial calculations on a local PC, local areanetwork or in a cloud environment without source code change.Quant can develop financial model using SixFaceCloud framework (Java or DSL) and execute it inrequired Execution Engine context: locally, in local network cluster, on Amazon EC2 and on GPU(coming soon).The user just needs to change few lines in a configuration file to offload the execution of a function tolocal network or cloud based servers. SixFaceCloud will deploy all the code to the executionenvironment and run it for you. Getting on the cloud has never been this easy!SixFaceCloud improves the full cycle of software development and deployment. No additionaldeployment steps are required in order to run your software on a cloud environment.SixFaceCloud enables you to develop faster, easier, and smarter.System StructureSixFaceCloud consists of 2 parts:• Development tools: Java framework and SFC Domain Specific Language(DSL). Tools are designed for rapid development of quantitative models (based on Monte Carlo methods for example) and for communicating with Execution Engine.• Execution Engine which can be installed in the local network or in a data center. It is designed for distributed execution of financial models created within SixFaceCloud Java frame- work or SixFaceCloud DSL. Download Execution Engine.Below is a short instruction on how to configure Execution Engine for local clusterSixFaceCloud infrastructure contains a single cloud server and multiple cloud workers. Adding moreworker nodes to the system increases computing power improving performance of Monte Carlosimulations. SixFaceCloud provides server and workers service libraries: cloud_server and cloud_worker. © Copyright 2012 SixFaceCloud, Inc. 1
  2. 2. Cloud Communication data : - Server static IP - Client connection port - Workers connection port Cloud client Cloud server Cloud workersNetwork requirement: server must have static IP address!Note that cloud server and cloud workers should be in the same network and client computer shouldbe able to establish TCP connection with cloud server (i.e.: client computer can be in another network).System ConfigurationServer’s IP address must be defined in configuration files of the cloud server and workers config/cloud_server/spring_cloud_server.xml property cloud.address config/cloud_worker/spring_cloud_worker.xml property cloud.addressBelow is a code snippet from configuration file that needs to be modified.Substitute YOUR_SERVER_HOST with your real server host address. <!-- Change YOUR_SERVER_HOST value to your real server host address. Note please, that it should not be 127.0.0.1 or localhost values. --> <bean id="cloud.address" class="java.lang.String" lazy-init="false"> <constructor-arg value="YOUR_SERVER_HOST:47500" /> </bean>Make sure that ports 5678, 5679, and 47500 are open and free on the server node. Please checkaccording policy if you are using Amazon EC2 hosting. If you cant use these ports, change serverPortand workerCommunicationPort in config/cloud_server/spring_cloud_server.xml file (see property‘comm.srv’). © Copyright 2012 SixFaceCloud, Inc. 2
  3. 3. System InstallationJust copy the content of delivery folder to a server and workers.Starting the System Run the server with server.cmd/sh (start server) Run every worker with worker.cmd/sh (start workers)Server must be started prior to workers. It is not necessary to restart already running workers if a serveris turned off. However, new workers will not be able to start without a running server instance.Additional configuration parametersAdditional server settings:File: config/cloud_server/spring_cloud_server.xml, service: ‘comm.srv’: Parameter Default value DescriptioncloudClientPort 5678 Cloud server connection port. It uses to connect to the cloud from the client side.workerCommunicationPort 5679 Cloud server connection port. The same value should be set on server and worker sides.classRequestTimeout 5000 ms Distributed class-loader timeout.randomServiceTimeout 2000 ms Random services communication timeout between server and workers.debugStatistic False Turn on/off statistic service debug mode.calcOnCoordinatorSide False Identifies if a server node takes part in calculation.corePoolSize 1 Parameter of the cloud server thread pool executor that is designed to communicate between cloud server, and workers, cloud clients.execAttempts 3 Number of execution attempts if user task is rejected from cloud server thread pool executor.execErrorWaitTime 500 ms Time in milliseconds before next execution attempt if user task is rejected from cloud server thread pool executor.heartbeatTime 3000 ms Heartbeat interval between cloud sever and cloud client.keepAliveTime 0 Parameter of cloud server thread pool executor that is designed to communicate between cloud server and workers, cloud clients.maxBatchOnNode 50 Maximum number of batches on one node. If user defined value is greater than given value, batch size should be increased, but batch quantity - decreased. © Copyright 2012 SixFaceCloud, Inc. 3
  4. 4. maximumPoolSize 200 Parameter of cloud server thread pool executor that is designed to communicate between cloud server and workers, cloud clients.maxMemoryUsage 67108864 bytes Maximum size in bytes of an object’s summary that can be accumulated in user code.maxObjects 100000 Maximum number of objects that can be accumulating in user code.maxTaskTime 60000 ms Maximum task execution time. User defined task execution time can be smaller, but not greater than this value.missedHeartbeatCount 3 Maximum number of missed heartbeats between cloud server and cloud client. The connection will be forced to close after this number is reached.serverSocketReconnectTime 500 ms Sleep time between server sockets creation after IO errors.disconnectTime 500 ms Sleep time before closing client connection.disconnectTime 500 ms Sleep time before closing cloud client connection.File: config/cloud_server/spring_cloud_server.xml, service: ‘stat.srv’ Parameter Default value DescriptionturnedOn True Turn on/off statistic service. © Copyright 2012 SixFaceCloud, Inc. 4
  5. 5. Additional workers settings:File: config/cloud_worker/spring_cloud_worker.xml, service: ’comm.srv’ Parameter Default value DescriptionworkerCommunicationPort 5679 Communication connection port between server and workers. The same value should be set on server and workers side.classRequestTimeout 5000 ms Distributed class-loader timeout.randomServiceTimeout 2000 ms Random services communication timeout between server and workers.debugStatistic False Turn on/off statistic service debug mode.calcOnCoordinatorSide False Identifies if a server node takes part in calculation. © Copyright 2012 SixFaceCloud, Inc. 5

×