The solution must have controllable isolation level of transactions and it must showcase the phenomenon of Non-repeatable reads or Phantom reads
Must use a true relational database or object-relational database
The database must consist of at least 5 tables including junction tables, 2 views and 1 M:N relation
The attributes of tables must use the following data types at least once
- Real number (float)
- Logical value (bool, ...)
- enum
- Text string (varchar, text, ...)
- Date or time (date, timestamp, ...)
Must allow creation, updates, deletion and display of a resource which splits into more tables at database level
Implement at least one transaction between at least 2 tables
Ability to generate reports which contains aggregated data from at least 3 tables
Ability to import data to at least 2 tables in format CSV, XML or JSON
Configurability via configuration file
Proper error handling for all edge cases

In addition to these criteria, there are the following extras:

1x Documentation
- This document must conform to the Documentation Checklist
3x Test cases
- All test cases are in PDF format
- 1 Test case is for running the app including importing database structure
- 2 Test cases are for testing the app itself and all possible errors and edge cases and configurability

Solution

My solution will use the following tech stack:

The problem in question will be demonstrated on a library application

lambda