← Back to context

Comment by spockz

2 days ago

IME, the suffering of bad performing code is mostly secondary. It increases compute costs. Mostly because requiring more beefy VMs than strictly required which is still benign and possibly more cost-efficient than spending more engineering effort. Sometimes because of the lack of performance now more scaling and orchestration is required which comes additional complexity and therefore compute and staffing cost. This is rare to get noticed and fixed due to organisational momentum.

The worst is when the performance is so bad it starts to prevent onboarding new features or customers.

The real cost is an opportunity cost. It doesn't show up in the financials. Your ability to react quickly to new business opportunities is hurt. Most CEOs and boards don't notice it, until it's too late.