CTOs Warn of Misguided IT Metrics and Software Performance Pitfalls

In a recent InformationWeek podcast, Jerry Shu, CTO and co‑founder of Daylit, and Aaron Harris, CTO of Sage, discussed how internal metrics can steer IT projects toward undesirable outcomes. They emphasized the need for broader organizational input when designing metrics, the importance of contextualizing data, and strategies for course‑correcting when optimization efforts fail to improve operations.

A complementary technical guide highlighted common reasons why software performance on modern Linux servers can degrade despite prior optimization work. It advises checking that the correct binary is running, verifying linked dynamic libraries, managing CPU affinity, and accounting for NUMA effects to ensure workloads stay close to the resources they use. The guide provides concrete Linux commands such as md5sum, ldd, taskset, and numactl to diagnose and resolve these issues.