And yet, despite decades of exponential growth in processing power, performance still matters, and in many ways, more than ever. Modern software runs at vast scale, processes extraordinary volumes of data, and increasingly lives in data centres that consume a growing share of the world’s energy. Inefficiency today doesn’t just cost money; it consumes power, adds operational risk, and ultimately has an environmental impact.
There’s a well‑known saying, often jokingly referred to as the fundamental theorem of software engineering:
“Any problem in computer science can be solved with another level of indirection.”
This is commonly followed by its less-famous corollary:
“Any performance problem can be solved by removing a layer of indirection.”
Abstraction and indirection are powerful tools. They enable reuse, flexibility, and faster change. But they are not free. Each layer adds overhead, complexity, and distance from the underlying data and platform.
When AMS built the technology platform that was used to build AMS payroll functionality, maximum efficiency and throughput were explicit design goals. A key contributor to its performance was a deliberate decision to minimise unnecessary layers of indirection. That came with trade-offs. The foundation layers of the platform, including the database, are more closely coupled than many modern systems.
The upside, however, has been significant. Because the platform is efficient, application complexity and data volumes are far less constrained in practice. The AMS payroll system, built on this strong platform, retains far more detailed historical data than many competing systems, without requiring archiving or consolidation, while still using considerably fewer hardware resources.
That efficiency delivers real customer value. During recent payroll remediation projects, payroll management customers had full historical data available - data that made certain remediation calculations possible that wouldn’t have been possible otherwise.
So, the question isn’t whether software efficiency still matters, but when we choose to prioritise it. Hardware may be cheaper than ever, but efficiency still buys us scale, sustainability, and long-term resilience, and that’s a trade-off worth thinking about.