This chapter covers the following topics:
Oracle Value Chain Planning Suite has a component architecture that separates the transaction data and processing, for example, inventory receipts and order entry, in a source instance and the planning calculations done in a destination instance. This results in:
Better system response
Planning calculations, for example, demand planning, inventory planning, supply planning and order promising, applying simultaneously to information from multiple source instances, which is useful for a global supply chain spread across multiple source instances.
The source can be any ERP system, but out-of-the-box integration to the Oracle Value Chain Planning Suite destination instance (planning server) exists in some cases but not in all cases.
Both source and destination database instances must be on the same major release of the Oracle database; see Source and Destination Instance Setup.
Oracle Value Chain Planning's data collection architecture, shown in the figure below, depicts the data objects, procedures, and data flow between source data and Oracle planning applications. The major repositories are ADS, ODS, and PDS. Procedures enable data cleansing, data collecting, data communication, and net-change handling between data repositories.
Oracle VCP supports the following planning configurations for installation and deployment:
These configurations offer you enough flexibility to design a mode of planning that suits your business objectives. Both configurations are supported using a consistent architecture as outlined in the previous section. The sole distinction is that centralized planning does not use database links to pull data into the Oracle VCP server data store.
To make more detailed machine deployment decisions, review :
This figure shows the centralized planning configuration.
In a centralized scenario, Oracle VCP and its source data reside in the same database. No database link is required in this case. Two components can communicate through the planning object APIs and the interface tables defined in Oracle Applications.
In this configuration, shown in the following figure, a simplified architecture is used, and the data transformation is not required.
When patching for a centralized installation, all patches are applied to the single instance. Ignore references in the patch readme for centralized and distributed installation and apply all patches to your single instance. Most patches mention to just apply the patch to your instance when centralized.
The following figure shows the decentralized planning configuration:
In this scenario, Oracle VCP works as a central Planning Server across one or more source data instances. The collection program is installed on the planning server and the data stripped by instance is moved into staging tables within Oracle VCP during the data collection process.
After the planning process, results can be pushed back to each instance. The planning process is generally done by a planner via the Planners Workbench.
Benefits of distributed installation:
The processing performed for Planning in ASCP, IO, DRP, etc. can take huge amounts of memory and disk IO. This can slow down the OLTP (OnLine Transaction Processing) processing while the plan processes are running. It is not unusual for large plans to consume anywhere from five to over twenty gigabytes of memory while running and move millions of rows in and out of tables to complete the planning processes.
It is possible to have VCP destination on a different release. VCP destination could be 12.2 with EBS source running 12.1. You may not have an VCP destination with a lower version than the EBS source. For example, VCP destination on 12.0 or 11.5 with EBS source on 12.1. For more information about compatible versions, please see My Support Oracle Note #1361221.1 -- FAQ - Getting Started With 12.2 for Value Chain Planning (VCP) Applications.
It is possible to have more than one EBS source connected to the VCP destination. This requires careful consideration as patching for all instances must be maintained. For example, two EBS source instances connected to one VCP destination instance and all are on the same EBS applications release. For VCP patches that affect collections, all three instances must be patched at the same time in order to prevent breaking this integration. Another example might be a customer with two EBS sources on different releases. VCP destination on 12.1 with an EBS source on 12.1 and another on 11.5.10.
Patching for a distributed installation requires careful attention to the patch readme to make sure the patches are applied to the correct instance and that prerequisite and post-requisite patches are applied to the correct instance. Patch readme should be clear on where the patch is to be applied. Usually, the patch refers to the EBS source instance and to the VCP destination instance where the patch is to be applied.
Consider from among these machine implementation options if you are implementing VCP products that use collected data other than Rapid Planning. For Rapid Planning, see Rapid Planning Machine Implementation Options.
Oracle Demand Planning also uses a third instance, an OLAP Analytic Workspace (AW) database, to hold data while multidimensional manipulation of demand data occurs.
For small implementations, source, destination, and Express can reside on the same machine and be in the same instance. The following figure illustrates this configuration.
Note: This example uses ASCP and DP. ASCP means Advanced Supply Chain Planning and DP means Demand Planning.
For larger implementations where system throughput is more important, the various instances can be deployed on separate machines. A two-machine deployment configuration is appropriate when the size of the demand planning data is relatively small. The following figure illustrates this configuration.
A three-machine deployment allows for the manipulation of high-dimensionality, large-scale demand planning data to occur on a machine separate from the planning calculations done on the planning server. The following figure illustrates this configuration.
The Advanced Supply Chain Planning concurrent manager may also be deployed on a separate machine. This creates even greater system performance. The following figure illustrates the four-machine implementation.
In all deployment configurations, a collection process brings data from the EBS source tables to the VCP destination tables. A build process brings data form the VCP destination tables to the OLAP Analytic Workspace (AW) for Demand Planning. Finally, a publish process moves data from the AW to the VCP Destination tables to be used for planning and can also publish data back to the EBS source instance.
Global Order Promising and Inventory Optimization planning calculations are also performed on the planning server.
Consider from among these machine implementation options if you are implementing Rapid Planning. For VCP products that use collected data other than Rapid Planning, see General VCP Products Machine Implementation Options.
Rapid Planning has these components:
Source ERP application, for example Oracle e-Business, JD Edwards, and legacy systems: A database tier and an application tier.
Value Chain Planning planning server: A database tier and an application tier. The database tier supports and manages the Oracle database instance. The application tier includes forms services, web services, and the concurrent processing server.
WebLogic Server
All the components are on one machine. The Source ERP application and the Value Chain Planning planning server use the same database tier and application tier.
Use this option with these data volumes:
Items: Fewer than 10,000
Demands: Fewer than 50,000
Planning users: 1 to 5
If you use this configuration with higher data volumes, you can overload your server.
The source ERP application is on one machine. The Value Chain Planning planning server and the WebLogic Server are on one machine.
This is how the machines communicate for collections and release when the source ERP application is:
Oracle e-Business Suite: Create a database link.
JDEdwards 9.0: Use the the JDE-VCP integration Process Integration Pack.
Another system: Use flat files/legacy collections.
The source ERP application is on one machine, the Value Chain Planning planning server is on one machine, and the WebLogic Server is on one machine.
This is how the machines communicate for collections and release:
The same way as the Two Machine (Distributed) Architecture
Cross mount the location for log files and data files that the concurrent programs on the VCP planning server write with the WebLogic server.
See:
Hardware Architecture, Oracle Rapid Planning Installation Guide
Sizing Template