A database will recover to a usable state when started after an unexpected shutdown. While we can achieve it without a database as follows:

• Write the whole updated dataset to a file
• Call fsync on the new file
• Overwrite the old file by renaming the new file to the old file, which is guarenteed by the file-systems to be atomic.

But this is only acceptable with a tiny dataset. Databases can do incremental updates.

There are two distinct types of database queries: analytical (OLAP) and transactional (OLTP).

• Analytical (OLAP) queries typically involve a large amount of data, with aggregation, grouping or join operations
• In contrast, transactional (OLTP) queries usually only touch a small amount of indexed data. The most common types of queres are indexed point queries and indexed range queries.

Data structures that persist on disk to look up data are called indexes in database systems. And database indexes can be larger than memory.

Common data structures include B-Trees and LSM-Trees

Modern applications do not do everything sequentially, and nor do databases. There are different levels to concurrency :

• Concurrency between readers
• Concurrency between readers and writers
• Do writers need exclusive access to the database ?