GumpDB

"He must be the stupidest son of a bitch alive, but he sure is fast!"

Idea

GumpDB works similar to cloakrooms, where you leave your clothes and in exchange they give you a number to later come back and fetch them.

GumpDB stores your records (C data structures) in a binary file and returns an ID each time you store a record.

With that ID you can then look up the values of the record.

Operations

Current

• Insert a record - O(n)
• Retrieves a record by id (doesn't take it out) - O(1)
• Delete a record - O(1)
• List records - O(n)
• Modify a record - O(1)

Locks

For simplicity's sake, Insert, Delete and Modify operations lock the whole file with an exclusive lock. And Retrieve and List operations place a shared lock on the file.

In this way, GumpDB enables many processes accesing records for reading while no other process is changing its content.

Maybe, think about, in the future:

Insert, Modify and Delete operations will lock its particular record bytes with an exclusive lock in the file. Retrieve will set a shared lock on its particular record bytes. And List operation will set a shared lock on the whole file.