9.2. A Simple Example
Consider a small toy store that sells wooden toys. The company decided to go online and set up its site a year ago. Because of its reduced customer base, management had a limited budget to develop the online project. As a result, the site came up with a very simple configuration. The e-commerce server setup consists of a Web server and a DB server that provide both the catalog content and the order forms. The technique used to connect the Web server and the application logic is Common Gateway Interface (CGI).
The following business functions are available at the site of the toy store: Welcome, Search, Browse, Product Information, Shopping Cart, and Checkout. By examining the logs of the site, we found that Search constitutes approximately 85% of all transactions. Thus, to simplify the problem, we assume that customers have only one navigational pattern: they visit the home page and then do a number of catalog searches for their favorite toys. The CBMG (Customer Behavior Model Graph) that models the navigational pattern and the CSID (Client/Server Interaction Diagram) of the search function are displayed in Fig. 9.1. It is worth recalling that the CSID describes the logical architecture of e-business functions, which are implemented by software entities called servers. In the CSID of Fig. 9.1, the search function is implemented by the Web and database servers, which run on a single machine.
Figure 9.1. CBMG and CSID Models for the Customer of the Online Toy Store.
Due to an international weekly magazine's article that described its unique handmade toys, the store's site is attracting the attention of online customers from all over the world. The company managers learned from others that doing business on the Web means dealing with potentially high traffic volumes at highly unpredictable times. Thus, they want to be proactive. In particular, management wants to know what the response time would be if the number of requests during the peak hour grew by a factor of four. To answer this what-if question, a performance model will be developed to represent the system behavior during the peak hour. Table 9.1 summarizes information regarding the system usage, collected during the peak hour.