A suggestion for an in-memory database, following up on Richard’s experiment

[jonathandlynch at gmail.com]

This is a different idea from the other thread, so I am starting a new thread. 

Imagine the following scenario:

Each record is saved as a separate text file.

LC loads up all the text files into a single massive global array.

LC listens to a port for DB requests.

When a DB request comes in, LC makes changes in the global array and retrieves data from the global array. It sends the results back almost instantly.

LC sends the new version of the file over to Apache which is listening on a different port.

Apache spawns a new thread for each concurrent request. Each thread simply takes the record and saves it to its file.

In this way, LC operates as an in-memory database, which is supposed to be very fast, and Apache does the multithreading to back up each record in the background.

Assuming a powerful server with lots of RAM, could this allow us to handle massive concurrency while using LC as the server?

Apologies if these questions are getting tedious - they are relevant to my current project.

Sent from my iPhone