Partitioning Application Logic on Servers

The logic for EnterpriseOne applications can be partitioned to run remotely by mapping individual or specified groups of business function components to run on an application server or enterprise server instead of on a workstation.

It has been found that redeploying certain business function components (including master business functions and business functions) can significantly increase the performance of a distributed EnterpriseOne workstation while simultaneously decreasing network traffic. This redeployment involves remapping objects using the EnterpriseOne standard Object Configuration Manager methodology.

Examples of such configurations are illustrated by the Windows light client/heavy server and the Java light client/heavy server models. Both models have applications specifications on the client and business function components on the server. The main difference is that the Windows model uses JDENet communication middleware, while the Java model uses JDENet/CORBA middleware.

While the software design enables you to partition all business function components, the biggest benefit is derived from partitioning Master Business Functions (MBFs).

EnterpriseOne transaction-oriented applications are built around the concept of MBFs, which are typically responsible for transaction edits and for committing transactions to the database. Most of the I/O services for transaction-oriented applications are performed by MBFs. By localizing the majority of business logic for transactions in MBFs and partitioning the MBFs to run on application servers, network traffic can be minimized, thus dramatically improving the performance of the application in distributed and WAN environments.

In a two-tier setup where MBFs are processed on the client, a lot of interaction occurs across the WAN between client and server. In a three-tier setup consisting of a client, a data server, and an application or enterprise server, transaction processing can occur across a LAN between the two servers. Interaction across the WAN between the client and server is thus reduced to entering input on the client and sending back results from the server. This three-tier configuration can result in a significant reduction in traffic across the WAN.

The following contrasts typical network traffic for a two-tier setup where MBFs are processed on the client versus network traffic segmentation for a three-tier setup where MBFs are processed on the server.