PL/pgSQL - An Introduction on Using Imperative Programming in PostgreSQL
Download as PPTX, PDF1 like1,949 views
The document introduces PL/pgSQL, explaining its importance in improving database interactions by allowing more complex actions through a procedural language rather than basic SQL. It covers features like stored procedures, triggers, and code blocks, which enhance performance and control over data consistency. Overall, PL/pgSQL provides a powerful, secure, and efficient means to handle dynamic database operations while minimizing unnecessary data transfer.
PL/pgSQL - An Introduction on Using Imperative Programming in PostgreSQL
- 1. PL/PGSQL AN INTRODUCTION ON HOW TO USE IMPERATIVE PROGRAMMING IN POSTGRESQL
- 2. Robert Sosinski Founder & Engineering Fellow
- 3. AGENDA PL/pgSQL: what it is and why it matters Volatility: living in an uncertain data world Upserts: a little update with a little insert Triggers: keeping data consistent Code Blocks: control when you need it Summary: bringing it all together Questions: fire away
- 5. HOW PL MEETS SQL PL/pgSQL SQL
- 6. WHY SQL IS NOT ENOUGH With SQL, every statement must be executed individually by the database server 1. Complex actions can require multiple round-trips 2. Intermediate results have to be marshaled and transferred 3. Multiple clients must implement redundant logic 4. Secure information might require exposure 5. Indexes and common filters might need complex logic 6. Consistency can require client-side interaction
- 7. TWO WAYS TO PL/PGSQL Stored Procedure • Stored directly on the database server • Can be used in queries and indexes • Able to be activated upon trigger conditions • May take input parameters and return values Anonymous Code Block • Sent to and parsed by the database when needed • Able to write to the database but unable to return a value
- 8. MORE PL THEN JUST PGSQL PL/Perl PL/TCL PL/Python PL/Ruby PL/PHPPL/V8PL/Java
- 9. DIVING HEAD FIRST Secure access database for a fictitious company with high performance needs.
- 10. LETS WRITE SOME FUNCTIONS Open Postgres Terminal
- 11. DEALING WITH VOLATILITY Volatility Categories provide assumptions the the query planner on how to optimize a function call
- 12. IMMUTABLE Characteristics • Cannot modify the database • Should not read from the database (side effect free) • Similar inputs should always return the same output • Can be used when creating indexes • Database may replace a function call with a constant Examples • Our check_access_for_user function for user types • The lower function for string types
- 13. STABLE Characteristics • Cannot modify the database • Able to read from database • Similar inputs should always return the same output on a per statement basis • Can be used in index conditions, but not creation • Can give caching benefits when called multiple times Examples • The current_timestamp and now function
- 14. VOLATILE Characteristics • Default volatility classification if none is specified • Can read from and write to the database • Can return different data on each call • Can be used as trigger functions • Planner will not make any assumptions for optimization Examples • The random and timeofday functions • Sequence functions such as nextval and currval
- 15. LETS TRY TO WRITE SOME STABLE FUNCTIONS Open Postgres Terminal
- 16. PULLING THE TRIGGER Trigger Function • The function to be run when a triggered event happens • Written using PL/pgSQL or any other PL language • Has special arguments available for control flow • If returns NULL, the event is canceled, else it continues Trigger Definition • The event that causes the trigger to be run • Always attached directly to the table or view • Can be run BEFORE, AFTER or INSTEAD OF • Can be run run for EACH ROW or EACH STATEMENT
- 17. LETS PULL SOME TRIGGERS Open Postgres Terminal
- 18. ANONYMOUS CODE BLOCKS The Good • You do not have to store the function on the server • Can run arbitrary PL/pgSQL when it is needed • Good for admin functions stored in your application The Bad • Cannot take arguments, so must be injected by your app • Unable to return a value, but you can get feedback • Must be parsed each time they are used
- 19. LETS “DO” SOMETHING Open Postgres Terminal
- 20. SUMMARY Powerful: have direct imperative access to your data Easy: syntax is familiar and simple to use Fast: because you are never leaving the database Secure: you do not need to send data back and forth Flexible: store the logic on the database or in your app
- 21. CONDENSED SUMMARY Direct access to your data means better control of your business