Most of the distributed databases offer consistency models, mainly strong and eventual consistency. Azure offers three more consistency levels. They are:

• Bounded staleness consistency: This guarantees that the read may lag behind the write. This consistency model increases the cost of a read option and is higher than session and eventual consistency.
• Session consistency: This is ideal for all scenarios in which a user is involved since it needs real-time reads and writes.
• Consistency prefix: This guarantees that the data written in order will be seen and read in the order that it was performed.

Let's look over the main promise of Azure Cosmos DB, which is its money-back guarantee.