4 ms·
On Scalability, Capacity, and Sensitivity
- jedberg 11y agoThis article is good, but there is one thing I take issue with: > For normal operation, you want to be within the engineered capacity of the system. But emergencies do happen and you get close to the peak of the system. If you've built a sufficiently monitored and auto-scaled system, then you're better off running close to the peak capacity at all times. You'll save money and find bottlenecks sooner.