Infrastructure is like an Ocean

Infrastructure is like an ocean.

At the edge of the ocean is land, the source of everything that is not native to the ocean; the users. Land and ocean; opposing duals.

On the surface it’s calm and steady. Maybe. A raft can survive on a good day. On a bad day being unsinkable won’t save you. Maintaining at this level is trivial if one knows the basics.

Below the surface, it’s different: There’s less light, less air, less… to say there’s less frames it in human terms. There’s more from the fish’s perspective; more pressure too.

Pressure and buoyancy, these are the keys to this metaphor. Near the surface and there’s little pressure and buoyancy is to be maintained at all costs; in the depths and consistency matters more than the absolute. Rapid changes in pressure are fatal. In infrastructure it’s important to know and keep steady all core system metrics to keep the system in specification.

In systems, pressure and buoyancy turn into two sides of the error rate. There are two tools at one’s disposal: serve different or serve less. Both require quantifying the error rate and necessitate a mechanism for doing so.

When architecting, think about where you’d rather be when errors hit.