Northern State University started encrypting 200,000 student records in January. By September, they had rolled back the entire system. Tomas Bergström, the database administrator who led the project, documented every mistake for future attempts.
Initial Planning Gaps
The team selected encryption software without testing it on their 15-year-old database architecture. Compatibility issues emerged immediately. Query performance dropped by 60%, turning 3-second searches into 12-second waits. Academic advisors who ran 50-80 queries daily faced hours of accumulated delay. The encryption keys were stored on the same server as encrypted data, creating a single point of failure that defeated the security purpose. Nobody had calculated the processing overhead, which required hardware upgrades the budget did not include.
Unexpected Benefits from the Attempt
The failed project revealed that 30% of their stored data was redundant or outdated. Cleanup efforts reduced storage needs by 60 terabytes. IT staff learned which database tables actually required encryption versus which needed only access controls. This distinction saved money in the revised plan. The university discovered their backup systems were inadequate when test restores failed repeatedly during the encryption rollout.
What They Changed
Bergström recommended a phased approach: encrypt only tables containing social security numbers and financial data first, then expand gradually. The revised budget increased from $40,000 to $95,000 to include necessary hardware. They scheduled the next attempt for summer break when system downtime would affect fewer users. The university now requires pilot testing on 5% of data before any full-scale encryption deployment.