So, how do you make database reads faster when you have a ton of data? That's the big question, right?

ive been runnin into systems lately where theres way more reading than writing. we did some basic caching, but its not really enough now. im wondering what cool tricks other engineers use to keep read speeds fast when dealing with millions of requests every second. do you guys use more read replicas, databases built for reading, or maybe somethin else altogether? lets chat