A scalable stock Web service

We built an Internet stock service and focused on its scaling characteristics. This service allows a person to request email notification when a specified stock reaches a price. For example, one could request to be emailed when IBM reaches or exceeds $150. We decided to leverage the capabilities of an underlying publication/subscription (pub/sub) system which has a JMS interface. We explored ways to configure the pub/sub system as well as how to tie it to a Web application server. We also explored performance results on different machines. Our goal was to determine what was necessary to scale to 3 million user stock requests as well as to have 100,000 satisfied each day and 100,000 new requests entered each day. We proved this possible