Supporting a vast educational empire in an online age often boils down to one major requirement. "We need to provide the best availability to our customers that we possibly can," says Apollo Group's senior director of software David Pinkus. "Our CIO is so committed to this goal that our bonuses are based on our availability from the user's perspective."
Apollo Group, parent of the University of Phoenix, has served working adults with extended learning programs for more than 30 years. The company's Web site (www.phoenix.edu) offers not only transaction systems that support enrollment but also message boards and access to student records and financial aid systems. "We have about 200 locations and all the systems [exchange] data," says Pinkus. "We need to know what's happening on which system and where we're getting performance degradation."
Pinkus says that the only way for Apollo to get true business-IT alignment is to focus on the user's experience. "Instead of saying 'server availability was X%,' we talk about how the student enrollment service was available 93.3% of the time and gave an average response time of three seconds." To understand and manage availability, Apollo uses Mercury's SiteScope and Application Mapping. Sitescope performance monitoring technology is part of Mercury's application management suite; Application Mapping discovers and documents relationships between applications and the underlying IT infrastructure.
SiteScope can send out proxy users and then "listen" to the performance of an actual transaction. Along with addressing immediate problems, the data helps Apollo understand "when we hit our biggest ebbs and flows" in concurrent users and what infrastructure upgrades will deliver the "biggest bang for the buck" in terms of availability, says Pinkus.
Application Mapping gives Apollo a better understanding of the topology of its complex infrastructure, which eases IT governance in terms of auditing and management. "One area where we've really struggled is in sorting out configuration differences between test and production," says Pinkus. "[Now] we're able to map and identify the differences between our development environment and specific production systems, and we can do a better job of customizing our deployment." This is critical, he adds, because performance is degraded when the IT group has to crawl the entire production environment every day.
Apollo Group doesn't use service-level agreements (SLAs) because "there's a lot of variability in our environment that depends on human behavior, such as when students log in to the system in different time zones," says Pinkus, who asserts that SLAs usually result from bad communication and mistrust. If business and IT have an adversarial relationship, "you have more fundamental issues than anything an SLA can solve," he says. At Apollo, technology helps business and IT stay focused on the same thing: working together to deliver a superior user experience.
— David Stodde