When is partitioning and sharding necessary?

Question

Partitioning and sharding databases

I am tormented by a question for which I have not yet found an answer: where is the border that partitioning and sharding is necessary?
Let's say there is a table Tickets, in which there are about 12 fields
- 5 text and 7 numeric. When reaching what number of records, horizontal scaling is necessary? Or at what number of records, the search will sag: 1 million, 10 million, 100 million?

Answer 1

the problem is that the percent and memory do not know about the number of records, but they know about their loading
https://github.com/faker-ruby/faker
you do it - the project will take off - you will decide: change the type of base, for example

Answer 2

This limit is different for each case and usually it hangs somewhere around the moment when the size of the table leads to applied negative consequences for the business.
Depending on the nuances of a particular project, the right time to scale up may come at 10 million records. Or it may not come at all - which means that suffering in this regard will be a vivid example of premature optimization (and such cases are the majority).
In a well-designed database, to which well-formed queries are executed, the moment of approaching the "border" will not be sudden and will be caught in advance by DBA with the help of monitoring.