Imagine you are designing a B2B service that will serve millions of businesses. This service will have dozens of different microservices with their own data, which can contain millions of records. How do you design such a database? Why is sharding not always the answer? What other options are there for such an architectural solution? I'll tell you how we at Uspacy came to serve thousands of small databases instead of a few large ones, what we've encountered and what we plan to face)