Skip to main content
 

In July, ITS Infrastructure & Operations improved the ConnectCarolina infrastructure to meet the surge in student registration activity due to COVID-19 using a capacity on demand strategy.

Scotia Roopnarine
Scotia Roopnarine

With the pandemic causing significant changes to courses for the Fall 2020 semester, mainly the different offerings for courses, ITS anticipated an unusually high level of registration traffic of normal students, plus additional activity due to the first-year student course registration. ITS was concerned that the registration system’s response time would be slower than usual, said Scotia Roopnarine, Director of Applications Infrastructure for ITS.

Employed capacity on demand strategy

Infrastructure & Operations teams brainstormed and identified performance issues in the infrastructure components that support student registration. This analysis pinpointed specific application server components responsible for the performance issues. Infrastructure & Operations created a plan to address these issues within the limitations of existing hardware resources.

This capacity on demand plan, Roopnarine said, was comprised of two components, add 25% more compute capacity to the application servers and implement Oracle memory management structure for the Student Administration student database. The additional compute capacity would enable the application servers to process more concurrent load, thereby lowering the response time during registration sessions, he said. Meanwhile, the memory management structure would enable the database to process work more quickly.

Shortened registration waiting time

Members of the Infrastructure & Operations and Enterprise Applications’ Student Administration teams watched as student registration unfolded together in live time via Zoom. They observed that the system behavior was vastly improved — the additional compute capacity and improved database performance shortened registration waiting time, thereby enabling a better student experience. Registration pages loaded with an average time of 0.7 seconds, compared to 13.0 previously, nearly 20 times as fast, Roopnarine said. Similarly, student batch jobs went from a run time of more than two hours to under seven minutes, a nearly 20 times improvement.

Will use for future registration events

Given the success of the capacity on demand strategy employed during Fall registration, Roopnarine expects the strategy will be used for future registration events, Finance year-end processing and other major University events.

Comments are closed.