SOLID principles for distributed systems: how do they translate?

we're taking a fresh look at how solid principles apply when you're designing distributed systems at scale. single responsibility principle feels pretty straightforward: it maps directly to service boundaries. open/closed principle also makes sense in terms of stable, backward-compatible APIs. where it gets a bit trickier is substitution principle and dependency inversion. when you have services interacting over a network, maybe with different data stores or deployment models, how do you really enforce that a 'substitutable' service behaves identically without perfect contracts and observability? and for dependency inversion, are we just talking about interface-driven development on RPCs, or is there a deeper meaning in terms of how services depend on abstractions rather than concrete implementations across a distributed boundary? i'm curious about practical patterns folks use to bake these into their microservice architectures.