RDBMS architecture

RDBMS are almost always designed as client/server systems:

• The server is the process that does the hard work
  • Reads and writes the physical files containing the data.
  • Checks that a user may login, and checks privileges.
  • Handles inserts, updates, deletes.
  • Performs the selects, applies all possible optimizations.
  • Communicates with the clients: receives queries, returns results.
  
  
• The client is a process that issues the commands to do inserts, updates, deletes, selects.
  • The client uses the database through a connection to the server.
  • It never accesses the data directly.
  • The client may execute on the same machine, or connect via the net from some other machine.
  • The client may be a GUI, a command-line interface, or some process doing calculations.
  • Most RDBMS has a ready-made shell-like client allowing a user to work interactively with a database using SQL commands.
  • Programs can become clients by using APIs (Application Programming Interfaces) in the code to access the data.
  • Interfaces exist for C, C++, Perl, Python for most RDBMS systems.