The first step is to design a remote schema. It is unnecessary and inefficient for each salesperson to have a copy of the entire consolidated database. The remote schema is designed so that it only contains information relevant to one particular salesperson. To achieve this, the remote schema is designed in the following way:

Each salesperson needs to keep records of all customers and products, so that any product can be sold to any customer. This tutorial assumes that a salesperson always speaks the same language as the customer, so you do not need the PRODUCT_DESCRIPTIONS table. Each salesperson needs information about orders, but not orders related to other salespeople. This is achieved by filtering rows based on salesperson identifier.

The next step is to choose the synchronization direction of each table. You should consider what information a remote database needs to read and what information a remote database needs to create, change, or remove. In this example, a specific salesperson needs a list of products and customers, but never entered a new product into the system. You are making the restriction that products and customers always enter the system from the consolidated database at headquarters. However, a salesperson needs to be able to record new orders on a regular basis. These factors lead to the following decisions about the synchronization in each table: