Women’s clothing brand Roman Originals has been inundated by calls since the photo became the center of an online debate.
Customize the content, goes the thinking at CheckOut.com, and shoppers will load their carts with videos and DVDs to do just what the name suggests. That’s why the entertainment site wants to tailor every page to fit customers’ profiles-and serve up to one million dynamically generated pages per hour. “The goal was to do that without buying a lot of hardware or enough Oracle to make Larry Ellison really happy,” says chief technology officer Edmond Mesrobian.
So many personalized pages, of course, threatened to slow down the site’s performance. Mesrobian, former vice president of technology at Disney Online, was well aware of studies showing that consumers waiting for pages to load run out of patience after only eight seconds.
CheckOut’s solution is a homegrown caching infrastructure that allows the site to access customer profiles without continually tapping its back-end database. It uses software developed in-house to query a front-end relational database that holds customer profiles, click-stream data and shopping carts. Only if the information being sought isn’t there does the system make the trek to get it. Connecting software keeps the two databases in sync, allowing more memory to be served from the site itself.
Traditionally caching appliances and software serve up static and commonly used content such as images and text. But systems like CheckOut’s are becoming more important to help sites create custom pages without bogging down their systems and driving away customers,says Peter Firstbrook, research analyst at the Meta Group, Stamford, Conn.
Times Ten Performance Software, a Hewlett-Packard spinoff that makes relational databases first used to speed up call routing and other tasks in the telecommunications industry, has taken the next step with a data cache called Front Tier. Check- Out, which uses a TimesTen relational database, consulted on the new app and expects to swap its data cache with Front Tier.