Mechanical sympathy is not bestowed by the abstraction — it is earned by studying what the abstraction hides ([[Cloud Abstractions Do Not Eliminate Hardware Variability]]). Kief Morris's prescription for a team running on a platform it doesn't own: - **Read everything.** Seek out every whitepaper, article, conference talk, and blog post about the platform you depend on. The vendor has documented how the physical layer behaves; most teams never look. - **Bring in the experts.** Have your vendor's engineers review your systems top to bottom — high-level architecture down to implementation — and ask explicitly how your designs interact with their physical infrastructure. - **Collaborate when it's yours.** If the virtualization layer is run by your own organization, there is no excuse not to design software and infrastructure holistically, then continuously measure and adjust for performance and reliability. The throughline: understanding the layer beneath you is an ongoing activity, not a one-time read. The same discipline applies wherever an operator depends on a system they did not build — a pilot studying the airframe, a trader learning the exchange's matching engine, a surgeon knowing the instrument. Expertise comes from interrogating the substrate, not trusting the interface. > "An infrastructure team should seek out and read through every whitepaper, article, conference talk and blog post they can find about the platform they're using." --- *Source: [External Link](https://infrastructure-as-code.com/posts/mechanical-sympathy.html)*