Demystifying Premature Optimization

This repo contains the full code of my blog entry.

Install instructions:

1. Run npm install. (This is a NodeJS app).
2. Create a file named .env in the root folder.
3. Having a instance of PostgreSQL 12+ running.
4. Create an empty database with a <some_name> you like.

The .env file content:

# Database settings.
DB_USER="postgres"        # Database login user.
DB_PASS="<some_password>" # Database password.
DB_NAME="<some_name>"     # Database name.
DB_HOST="localhost"       # Database ip.
DB_PORT=5432              # Database port.

# Behavior settings.
TABLE_SIZE=1.5e6    # How many total entries the tables will have.
SAMPLE_SIZE=0.667   # How many % of the entries will NOT be filtered.
ITERATIONS=3        # How many test iterations will be running per service version.
SETUP_DATA=true     # To check if the setup phase needs to be ran. 
#RECREATE_DATA=true # To re-create the data (Will override SETUP_DATA).
#LOG_QUERIES=true   # To print the queries sent to the database (some set-up data queries will be skipped).

Running instructions:

To be fair, the app node <file.js> must be ran using the flag --expose-gc, so we can use global.gc() to do a full garbage collector run before test, so the garbage collector does not run in the middle of the tests.

There are two 'different' ways of running it:

1. node --expose-gc index.js
2. node --jitless --expose-gc index.js

The way #2 actually disable the jit compiler with the flag --jitless. Please, be sure to disable RECREATE_DATA, if not your test may last a lot of time.

Notes:

Environment: SETUP_DATA --> This controls whether you want the setup phase to run, if this flag is set it will first check if the data exists, if not it will re-create the database, otherwise it will run the test using the pre-existing data.

Environment: RECREATE_DATA --> This will override SETUP_DATA and will skip the check, having this value as RECREATE_DATA=true will always re-create the data every time you run the tests.

Environment: SAMPLE_SIZE --> See setupDatabase.js for more implementation details.

Setup info:

• CPU: I7-9700K
• RAM: 32GB - 3200MHz
• Windows 10 (20H2)
• Node version: v16.5.0
• PostgreSQL: Debian 13.0-1 (Docker)

Running samples:

TABLE_SIZE: 1500000 | SAMPLE_SIZE: 0.667 | ITERATIONS: 4

service version	run 1	run 2	run 3	run 4
v1	4915ms	4690ms	4668ms	4608ms
v2	3593ms	3556ms	3598ms	3607ms
v3	2762ms	2736ms	2720ms	2720ms
v4	2515ms	2457ms	2479ms	2479ms
v4.1	1578ms	1530ms	1533ms	1541ms
v5	1723ms	1727ms	1798ms	1789ms
v6	1715ms	1720ms	1736ms	1730ms

For more samples you can go to /logs/* and read the .txt files. Also you can ran the test by yourself 😊.

Author:

Enmanuel Reynoso | DarkRoku12 | enmarey2012@hotmail.com