The framework and characteristics of IBM SPSS analytical Decision Management cluster environment

This article describes in detail how to deploy and configure ibm®spss®collaboration and deployment Services in a clustered environment. Ibm®spss®collaboration and Deployment Services Repository can be deployed not only on a stand-alone environment, but also on cluster application servers, where the same application is deployed on each application server in a clustered environment. A load balancer is used to access the storage repository.

Figure 1 is the architecture of the distributed environment, including application servers, storage libraries, load balancers. Logically, the architecture of distributed environment includes the basic concepts of unit, node and so on. Each cell is a set of nodes in a single admin domain, one unit contains one node or multiple nodes, each node can have multiple servers, but each server's configuration file is stored and maintained separately. In a distributed environment, multiple nodes in each unit are managed uniformly, and their configuration files are centered into the cell master configuration file. The Configuration Manager thread is responsible for managing these centralized configuration files and synchronizing any configuration changes to each node. A node is a logical group of server threads that can only have one node per machine and communicate with the Configuration Manager through the node agent. In the WebSphere ND environment, was plugin plays a load-balanced role in CFG files that define the weights of each server, depending on the weight value, plugin decide which server to direct a request to.

Figure 1. The architecture of the cluster environment

Through the cluster can realize more customers, improve throughput, through load balancing, so that resources can be effectively utilized, if a node failure, does not affect customer business access, has a good recovery and compensation mechanism.

Build IBM SPSS Collaboration and deployment Services cluster environment

Deploying Ibm®spss®collaboration and Deployment Services Repository in a non-clustered environment is very simple, but some large enterprises, in order to achieve large data exchange, the cluster environment is a good choice. This section of this article describes in detail how to deploy and configure ibm®spss®collaboration and deployment Services in a clustered environment.

The installation of IBM SPSS collaboration and Deployment Services is generally divided into two parts: the first is to use IBM installation Manager to automatically download and configure the installation files to the local file system from the data source. The second is to use the IBM SPSS collaboration and Deployment Services revisit Tool to collaboration and deployment services Deployed to application servers and databases.

Pre-Installation Preparation

Before installing IBM SPSS collaboration and Deployment Services, you need to do some preparatory work, such as installing, configuring creator server, setting up the Java environment, and creating a database.

Install IBM WebSphere and create cluster profile

There are a number of creator servers that support the clustered environment, as illustrated in the example of the WebSphere receptacle deployment version. Before installing a clustered environment, you must determine the size of the cluster, that is, the Node quantity. This article takes a minimum of two nodes as an example to explain, for more than two nodes installed, the method is the same.

First, we must install WebSphere on all cluster nodes, and the installation method can refer to other articles on the developworks. Second, you need to create the WebSphere profile, which includes a manager profile and two node profiles. Users can use WebSphere's own WebSphere customization Toolbox tool to create profiles in a GUI, or you can use the following script to create and start profile. Open the command line interface for Windows and go to the bin directory under the WebSphere root directory:

Listing 1. Enter the bin directory code under the WebSphere root directory

CD C:\IBM\WebSphere\AppServer\bin

Using the Manageprofiles command to C:\IBM\WebSphere\AppServer\profileTemplates\management as a template, create a profile with the name Dmgr01 and place it in the C:\ Wasprofile\dmgr01 Position:

Listing 2. Code to create WebSphere Manager profile

Manageprofiles.bat-create–templatepath C:\IBM\WebSphere\AppServer\profileTemplates\management-profileName Dmgr01-profilepath C:\WASProfile\Dmgr01

Enter the bin directory of the profile you just created and start it with the Startmanager.bat script:

Listing 3. To start the code for WebSphere Manager profile

CD C:\WASProfile\Dmgr01\binstartManager.bat

Use the following script to create node profiles on both machines and add them to Manager Profile management:

Listing 4. Create a WebSphere Node profile and start code

CD C:\IBM\WebSphere\AppServer\binmanageprofiles-create-templatePath C:\IBM\WebSphere\AppServer \ Profiletemplates\managed-profilename Custom01-profilepath C:\WASPROFILE\CUSTOM01CD C:\WASProfile \custom01\ Binaddnode.bat <DMGRHOST>

Where <DMGRHost> needs to be replaced with the host name or IP address of the system on which the manager profile resides. Additionally, you need to install the WebSphere HTTP Server as the load balancer. Readers can refer to the IBM Developworks related articles on WebSphere.

To deploy IBM SPSS CADS to WebSphere cluster environments. First, we need to create a cluster and configure it through WebSphere's admin interface. Use Http://<Dmgr_system_ip>:9060/ibm/console to enter the WebSphere configuration management interface. Select Clusters->websphere Creator server clusters in the left select list.

Figure 2. Websphere Console Servers List

At this point, from the right side of the interface can be seen at this time there is no cluster, we click the New button to create a new cluster. WebSphere evokes a wizard that creates cluster, instructs us to enter the cluster name, and adds the cluster child nodes.

Figure 3. Create a new cluster name

Figure 4. To create a cluster member

After all the required child nodes have been added to the cluster, WebSphere will give a summary interface, and then click the Finish button to complete the cluster creation.

Figure 5. Complete cluster creation

Configuration Java Environment

The manufacturer of the JDK is determined by creator server, such as WebSphere Creator Server, which requires the use of IBM's JDK. JDK, JRE use version 1.6. Set the Java_home variable and add%java_home%\bin to the Path.

After you set up the Java environment variables, enter java–version in the command window to check that the Java environment is set up correctly.

Create a database

IBM SPSS CADS supports a variety of mainstream databases, such as DB2, SQL Server, Oracle, etc. This article takes DB2 9.5 as an example of how to create a database.

database parameters created by default are not sufficient, and users must set parameters such as UTF-8, buffer pool size, temporary page table size, and so on. For ease of use, the reader can copy the following script to a file and save it on a computer disk, such as in C:\myScprit.sql. This script creates the database name is Spsscds.

Listing 5. To create the SQL code for the database required by CADS

CREATE DATABASE Spsscds on C:\DB2\CDS using codeset UTF-8 territory US COLLATE using SYSTEM; CONNECT to Spsscds; CREATE bufferpool cds8k IMMEDIATE SIZE AUTOMATIC PAGESIZE 8 K; CREATE REGULAR tablespace cds8k PAGESIZE 8 K consolidator by AUTOMATIC STORAGE extentsize 8OVERHEAD 10.5 prefetchsize 8 Transferrate 0.14 bufferpool cds8k dropped TABLE RECOVERY on; COMMENT on tablespace cds8k is '; CREATE bufferpool cdstemp IMMEDIATE SIZE PAGESIZE K; CREATE SYSTEM Temporary tablespace cdstemp PAGESIZE K consolidator by AUTOMATIC storageextentsize-overhead 10.5 Transferrate 0.14 Bufferpool "Cdstemp"; COMMENT on tablespace cdstemp is '; CONNECT RESET; Then click the Start button to select the program to run Db2-db2copy1 (default)-command line Tools-command Window. The DB2 command Line window is now open. Then type the following command to create the database:

Listing 6. Code to create a database

DB2-TVF C:\myScript.sql The final need to authorize the database, still select from the program where the Start button, Db2-db2copy1 (default)-general administration Tools-control Center command, right-click the database you just created select ' authorities ... ' command, add ' db2admin ' user, and give all permissions. Finally, close and exit DB2 Task Center, at which time the database creation is complete.

Installing IBM SPSS CADS

CADS5 compared to earlier CADS, it uses IBM Install Manager for installation steps and is installed through repository.

The installation of IBM SPSS collaboration and Deployment Services is generally divided into two parts: the first is to use IBM installation Manager to automatically download and configure the installation files to the local file system from the data source. The second is to use the IBM SPSS collaboration and Deployment Services revisit Tool to collaboration and deployment services Deployed to application servers and databases.

If you start C&ds installation, you first need to verify that the version above IBM installation Manager1.6 is already installed on the user's operating system, and that IBM installation Manager's data source is configured correctly. If the C&DS installation media detects that IBM Installation Manager is not installed, it automatically starts the Installation Manager installation. If the C&DS installation media detects that the version of installation Manager is too low, it will automatically upgrade to 1.5. For Windows, for example, the detailed installation steps are as follows:

logs on to the operating system with the right user and runs installation Manager. In command line mode, you can run

Listing 7. To start the code for the IBM installation Manager command line interface

&LT;IBM installation Manager Home&gt;/eclipse/ibmim/tools/imcl–c in interface mode, can run

Listing 8. To start the code for the IBM installation Manager graphical interface

&LT;IBM Installation Manager Home&gt;/eclipse/ibmim.exe on the menu bar of Installation Manager, click File-&gt;preferences, configure Installation The correct data source required by IBM SPSS collaboration and Deployment Services.

Figure 6. Add IBM SPSS Collaboration and deployment Services repository source

Of course, you can specify the path in the form of file system, network, HTTP, and so on. Then go back to installation Manager's main interface and click the Install button. Then select IBM SPSS Collaboration and Deployment Services version 5.0 for installation.

Figure 7. Choose to install the product

Click the Next button to proceed to the next step. Read and accept license agreement. Click the Next button. On this page you can set up the installation path for IBM SPSS collaboration and Deployment Services. Here we use the default path to install. The installation of IBM SPSS collaboration and Deployment Services requires a new package group. The installation of IBM SPSS collaboration and deployment Services requires about 5G of hard disk space. If you are starting Installation manager for the first time, you will also need to set up a shared directory of Installation Manager to temporarily store files downloaded from the data source. On this page you can configure the components you want to install. The default component is both IBM SPSS collaboration and Deployment Services version 5.0. The next page shows the installation-related information, click the Install button and start the installation. The installation process starts with the following figure.

Figure 8. CADS installation Process

installation is complete, as shown below. IBM SPSS Collaboration and Deployment Services version 5.0 will be downloaded and installed to the user's local file system via IBM installation Manager. If you find problems during this process, you can refer to the SPSS collaboration and deployment Services Install Log positioning and identify the problem. Click the Finish button. The SPSS collaboration and deployment Services revisit Tool will start automatically. After the installation is complete, you need to configure the CADS.

Configure IBM SPSS CADS

After the IBM SPSS CADS installation is complete, it must be configured before it can be used. The configuration process mainly completes the following three tasks:

Create a database object for the content repository, such as creating a datasheet and writing the content. Create creator server resources, such as JMS queues, and deploy Java-ee jar packs to creator server. Configure encryption and security.

Double-click the configtool.bat batch file in <cads Foler>\bin to start the Configuration tool for CADS IBM SPSS collaboration and Deployment Services revisit Tool. Click the Next button to enter the creator server configuration interface. Select IBM WebSphere in the Creator server type Drop-down list and set up the manager profile created above in WebSphere profile directory, that is, Dmgr01. CADS automatically determines whether a single Server type or Cluster type is based on profile properties. In the URL prefix, fill in the URL that accesses the Load balancer.

Figure 9. Configuring the Application Server

Next click Next to enter the database configuration interface. Select IBM DB2 in the database type and fill in the following configuration entry with information such as host name, port, database name, username, and password.

Figure 10. Configuration database

Click the Next button, CADS will try to connect to your configured database, if the connection is unsuccessful, it will give the error message to know you correct the error, if the connection is successful, the Configuration tool will enter the Existing data configuration interface.

If the user chooses a database that is empty, the user does not need to configure the exiting data page. If the user chooses a database that has IBM SPSS collaboration and Deployment Services installed, users can choose to reuse previous database information or start a new installation.

Next is the KeyStore configuration page. Keystore is actually equivalent to a key used to encrypt information such as passwords stored in IBM SPSS collaboration and Deployment Services. Users can enter a keystore password to encrypt keystore.

Next is the configuration page for the admin account of IBM SPSS collaboration and Deployment Services. Users can configure the password for the admin account here.

The user can then choose to automate deployment or manual deployment. In an automated deployment, all resources will be deployed automatically by IBM SPSS collaboration and Deployment Services revisit Tool. If you choose to deploy manually, IBM SPSS collaboration and Deployment Services revisit Tool will generate the resources needed for deployment, but require users to manually deploy them to the application server.

The last page is revisit Summary. All the configuration of the user just now will be listed on this page. By clicking the Configure button, IBM SPSS collaboration and Deployment Services revisit Tool began working on the IBM SPSS collaboration and Deployment Services is deployed to user-set application servers and databases. If there is a problem in this process, the user can view the install log to locate.

Figure 11. Configuration Information Rollup interface

The configuration of IBM SPSS Business Analysis System in the cluster

After the IBM SPSS CADS is installed and configured, you can install the Modeler Adapter and Decision Management. Before you install, you must set the CADS folder to shared mode, and then modify the WebSphere variables on the console of WebSphere. For example, modify C:\qatest\cads to \\hostname\qatest\cads.

Figure 12. WebSphere Console Configuration

WEB Cluster Environment creation

You can choose to WebSphere8 HTTP Server or WebSphere7 HTTP server as your load balancer, install HTTP server by default, go to Dmgr console, and select Web Servers on the left Servers menu option, let's create a Web server.

Figure 13. WebServer Configuration 1

Figure 14. WebServer Configuration 2

After building the Web server, enter the Servers->web Servers configuration interface and click the Propagate Plug-in button to propagate the webserver you just configured.

Figure 15. WebServer Configuration 3

After the spread is complete, start WebServer1 and cluster.

Figure 16. WebServer Configuration 4

Figure 17. Start the Cluster interface

Once the configuration is complete, the user can install the Modeler Adapter and Decision Management on the CADS cluster environment. Installation is the same as a stand-alone version of the installation, users can refer to the IBM Developerworks IBM SPSS Decision Management Enterprise-level applications deployment articles.

Build IBM SPSS Modeler Server cluster environment

This article takes the cluster of two modeler servers as an example to explain the configuration of the Modeler server cluster environment. The Modeler Server installation method is very simple, just follow the installation interface Next down, this article will not repeat. Once installed, enter into the Modeler Server installation directory on both systems and delete the Options.cfg file

"Administrators," * "", and then restart the Modeler Server service.

Next, we need to configure these two Modeler servers with the Modeler Administrator Console to run them in a clustered way and work with the WEB cluster environment that was built up earlier.

Figure 18. Modeler Server Cluster configuration

When the configuration is complete, restart the WEB cluster environment.

Use IBM SPSS collaboration and Deployment Services Manager to configure the Modeler server in Repository. By opening IBM SPSS collaboration and Deployment Services Manager and logging into your environment, you will find that there are two Modeler server objects in Servers. Right-click Server clusters to create a new server Cluster Definition.

Figure 19. New Modeler Server Cluster name

Figure 20. Add existing Modeler Serve nodes to the cluster

Add the Credentials named Modeler to the Repository.

After the configuration is complete, enter Http://hostname:80/DM in the browser, and then you can start your cluster tour of IBM SPSS analytical Decision Management. When the system is running, it dynamically mixes which modeler Server to model according to the load situation.

Solution to common problems in Web cluster

After completing the above steps, the user should be able to open the IBM SPSS analytical Decision Management Environment by entering Http://hostname:80/DM in the browser, and if not, try the following series of check steps.

Log in to the WebSphere Admin interface: Enter Http://hostname:9060/ibm/console in the browser. Check that WebServer1 server is in the cell in WebSphere (System Administration-&gt; cell) to verify that HTTP server is propagated. Click the Generate Plug-in button in WebServer1 server (Servers-&gt; Web Servers) to see if the Web server was successfully created. Check the CADS installation path to see if CADS has been successfully installed. If Plugin-cfg.xml is not updated, try updating the file by first deleting the webserver and recreating it, then clicking on the "Generate Plug-in" button, at which point Plugin-cfg.xml can be updated. The WEB Cluster management interface appears similar to the following log:plgc0005i:plug-in revisit file = C:\IBM\WebSphere\AppServer\profiles\Dmgr01\config\ Cells\ Dg1dm-data2cell01\nodes\dg1dm-clst-2anode01\servers\webserver1\plugin-cfg.xml when &nbsp;generateplug-in When &nbsp; is finished, click the &nbsp;propagate plug-in&nbsp; button. This step automatically copies the updated Plugin-cfg.xml files from the Domain Manager directory to the HTTP server directory. The WEB Cluster management interface appears similar to the following Log:LGC0062I:The plug-in revisit file is propagated from C:\IBM\WebSphere\AppServer\profiles\ Dmgr01\config\cells\ Dg1dm-data2cell01\nodes\dg1dm-clst-2anode01\servers\webserver1\plugin-cfg.xml to C:\IBM\ websphere\httpserver\plugins/config/Webserver1/plugin-cfg.xml on the WEB server computer. Plgc0048i:the propagation of the plug-in revisit file is complete for the WEB server dg1dm-data2cell01 Complete these steps, click &nbsp;Start&nbsp; button to restart webserver. and restart the HTTP server in the Windows service.

Concluding

This paper takes Windows platform as an example to explain the deployment and configuration of IBM SPSS analytical Decision Management Commercial Analysis system in the cluster environment, and the process of installation and configuration on Linux and UNIX platform is basically the same as that of Windows. I hope this article can provide some reference and help for the application and deployment of IBM SPSS analytical Decision Management cluster.