Databases are stored on a server and it will handle transactions that are authorized to use it. The location of the database is called by the software, and in many cases, the server is then setup to deal with the amount of requests.
For instance, with Mongo, you have a cluster of shards. Shards being multiple occurrences of a database that work much like multiple IP address for a web site. They assist with the workload, and if one is burdened or goes down, the others are able to cover for it.
Then you have a CDN (Content Delivery Network). Similar to what was described above, they serve your assets to the user. Documents, images, etc. Even though you aren't directly storing assets within a database, the URLs or location of those files may be stored so when that data is pulled, an image element will then have it's source provided for it.
However, if you're working on creating a product that a small team will use internally - you won't be facing a flood of users. So a private CRM, used by a small company wouldn't have nearly as many chances of experiencing server issues like a CMS would (like a blog) if it was public and popular.
I hope that helps a bit!