|< Day Day Up >|
Several weeks have passed since the disastrous airfare sale described in Chapter 19, "Case Study: Friends Fly Free-for-All A Promotion Gone Wrong." The new CIO has settled in and begun a large-scale optimization project. You did such a good job on the previous two projects that you've been tapped to lead the task group. Surveying the user and technical groups tells you that things are better, but still far from perfect. Some of the old problems still remain, and some new issues have cropped up.
In particular, High-Hat Airways recently scrapped its partnerships with other airlines such as Germany's Kleine Sitzfluglinien, Spain's Vuelos Incómodos, and the French carrier LAD (Ligne Aérienne Désagréable). The old program has been replaced with a new group of partners. Being primarily composed of second-tier fast-food restaurants, dry cleaners, and laundromats, these new partners do not have the cachet of the original group, but they offered better financial terms. Their customers will be able to earn one High-Hat mile for every dollar spent in these establishments. However, because it takes at least 25,000 miles to earn a free trip, it will take decades for many of these new plan members to earn a trip. Nevertheless, the revamped program is a hit with a whole new constituency, many of whom have never even flown.
Within 30 days of the program's launch, massive data files containing information about the partners' customers, along with their mileage-earning purchases, began arriving. The first problems related to the files themselves; most were incorrectly formatted and were unable to be loaded. This caused no end of aggravation for the reservation agents, who had to field thousands of calls from irate new program members demanding to know why the few dozen miles they earned eating fast food during the last month aren't in their accounts yet.
Eventually, the data files are correctly formatted and successfully loaded each night. This introduces another, more intractable issue it simply takes too long to import this information.
You also learn that the clustering solution you recommended has had mixed results: Database server availability is much higher, but many users are complaining of terrible query performance.
Finally, you had counseled High-Hat to start using MySQL 5's stored procedures as a way of centralizing business logic while improving performance. The good news is that the referential integrity and other problems that had plagued the airline have largely disappeared. The bad news is that some of the stored procedures are perceived as running more slowly than the client-side application logic they replaced.
Other problems also exist, but because the users are complaining most loudly about the preceding three problems, you decide to make solving them your priority.
|< Day Day Up >|