Installation Process

Contents

• Introduction
• Analysis Models Installation Process
• General Considerations Related to IFS BI Analysis Package Installer
• Code Modifications Scenarios
  • Scenario 1 - Standard Code Modifications
  • Scenario 2 - Customized and Standard Code

Introduction

This section called 'Analysis Models - Support and Upgrade Process' gives information and guidelines related to the support and upgrade process of the Analysis Models.

Note: The Analysis Models were formerly known as IFS BI Analysis Package.

Note: Related example content in the Analysis Models and IFS BI Analysis Package, such as SSRS and Power BI reports, are not supported.

Analysis Models Installation Process

Suggested Installation Process

Click image to show full size

• Modifications are always performed in the customer code base. Even if there are no customizations involved, i.e. only core corrections involved, it is important to update the code repository.
  
• New code means that creating a new IFS Build Home is the preferred action. It is not recommended to overwrite an old build home.
  
• Next consideration is if the corrections should be applied to a test or a production environment.
  
• The process is nearly the same for test and production:
  
  • Start with rebuilding the IFS BI Instance Home.
    A current instance home can be overwritten as long as the user handling the installation knows what the consequences are.
    The other and preferred option is to rebuild the instance home by simply creating a new one that still has the same target environment as the old one. These instance homes can now be examined to find out differences.
    
  • Should the modifications be deployed manually?
    • If yes, and this is considered as the main alternative, the modifications must be tracked, all files have to be manually modified, database modifications as e.g. table properties and data must be planned and also discussed with the customer in order to have an idea of accepted actions.
    • If no, then the Analysis Models Installer can be used to install everything, i.e. building all files, performing deploy actions and also create one SSIS specific Agent Job that can be used to run the ETL process.
      Note however that this option is not likely to happen rather often since a complete rebuild means recreating everything, i.e. also reprocessing the data.
      
  • Run the ETL related packages in order to update the Data Warehouse. This can be done by using the IFS provided method of running a command script that starts an SSIS package which in turn runs the ETL process. It might however be necessary to run only a few sources in one go, setting up different schedules etc.
    
  • Verify the corrections/modifications.
    
  • Finalize:
    
    • If the test environment was the target environment, then it is important to make sure that all corrections are reflected in the code base in the code repository. It should be possible to extract the code and build an exact copy of the BI instance. The code in the repository is then used to build a new BI Instance Home for the production environment.
      
    • If the production environment was the target environment and verification fails, it is of course possible to figure out what is wrong and to do a quick fix. Another possibility it to get the previous verified version of the code base and use it to get the environment up and running.
      

General Considerations Related to the Analysis Models Installer

A general note and recommendation:
It is NOT recommended to run the deploy steps of the installer for SSIS files, SSAS files or Database Components in a support or upgrade scenario. SSRS reports can be automatically deployed however but in this case note that existing files with the same name will be overwritten without a warning.

The Analysis Models Installer can be used for two main purposes.

• Step 1: Create BI Instance Home
• Step 2: Deploy to target SQL Server(s)

The first step creates the BI Instance Home in a destination folder decided by the user running the installer. The BI Instance Home will contain database files, SSIS-packages, SSAS (OLAP-cube) files and SSRS-reports. The BI Instance Home is based on a specific IFS Build Home that contains the consolidated files revisions from all involved BI components. This step should always be run if a change or new functionality has been added to the code base.
The second step deploys the files in the BI Instance Home to the target SQL Server(s). The deployment step has different principles, SQL Server deployment and file structure deployment, for different types of SQL Server components.

• SSIS-packages are executed from the BI Instance Home directory. This means that if a new BI Instance Home is created then the SQLServerAgent job running the packages must be reconfigured to point at the new Instance Home.
• Database files (tables, views, stored procedures) are in a fresh installation deployed in the SQL Server database. Modifications to files creating database objects will be executed but no modifications will be made since the database objects are not replaced. This means that modifications to database files must be handled manually.
  There are two ways of handling this:
  1. Manually drop the old object(s) in the SQL Server database. This means that all existing data in a table disappears and must be re-generated. Execute the new SQL-script to get the new version of the object into the database. This option is most suitable for views, indexes and stored procedures. For tables it is a bit more complicated due to risk of data loss.
  2. Open the existing object and manually edit it to include the changes. This option is most suitable for Dm_* tables if data needs to remain in the tables after the update.
• SSAS-files are not automatically deployed to the database if re-executed. Changes to SSAS-files should be entered manually using SQL Server Data Tools.
• SSRS-reports are deployed to the Report Server. If a report with the same name already exists on the report server it will be overwritten with the new version if the checkbox Overwrite Existing Definitions is selected.

Code Modifications Scenarios

Per definition a complete build, from an IFS perspective, means a build where the Analysis Models installer is executed from an IFS Build Home. Databases, tables etc. will be deployed to the SQL Server, an Agent Job is created which will be used to start the ETL process in order to transfer data from IFS Applications to SQL Server.

Scenario 1 - Standard Code Modifications

In this scenario there are no customizations. Bug corrections from IFS only affect standard code.

From an installation perspective it is possible to take the complete Analysis Models code after a bug correction and redo the installation through the Analysis Models Installer. But since there is no support for incremental updates or single patch modifications, a rebuilt BI Instance Home is useful for an installation ONLY if it is possible to totally recreate everything. If data in the test or production environment should be kept, then the update/modification handling must be given special considerations.

Modifications may need to be merged manually and deployed from the BI Instance Home to the production environment, if for example the ETL load or cube load time is unacceptable for a complete re-build. This could require table changes and migration of data etc in existing installations. Note that the customer most likely would like to keep the data in the Data Warehouse and get patches added without affecting the production environment in any negative way. So the modifications/additions will have to be planned carefully.

Necessary information about what to deploy and modify should be found in the Single Patch documentation attached to the LCS bug correction.

Scenario 2 - Customized and Standard Code Modifications

For customized code it is recommended to keep all modifications in customer specific versions of the IFS BI components.

A hint for what to consider for the merge can be found by looking at the Single Patch documentation, but the recommendation is to perform a diff (manual) between the existing files in the installation code base and the patch and then take actions according to found differences.

Modifications may need to be merged manually and deployed from the BI Instance Home to the production environment, if for example the ETL load or cube load time is unacceptable for a complete re-build. This could require table changes and migration of data etc in existing installations. Note that the customer most likely would like to keep the data in the Data Warehouse and get patches added without affecting the production environment in any negative way. So the modifications/additions will have to be planned carefully.

Please note that only core code is supported by IFS Product Development. Customized code is not supported.