'Optimizing for reads' becomes a premature optimization when you're spending time making something faster that doesn't actually need to be faster yet.

Everyone talks about making databases good for lots of reads, you know, with denormalization, read replicas, and all sorts of caching. It just seems like the go-to advice in a lot of system design chats. But when does all that focus on optimizing for reads, especially right at the beginning, become just too soon? If your app isnt getting much traffic yet, or if how you write data is actually way more complicated than you thought, could all that read optimization just make things harder than they need to be? I'm wondering about times people have really struggled to figure out that balance.