A snowflake algorithm implemented as Postgres Extension. It uses time as first component and remain sortable. Ids are 64-bits, half the size of a UUID.
ID is composed of:
- time - 41 bits (millisecond precision w/ a custom epoch gives us 69 years)
- instance id - 10 bits - gives us up to 1024 postgres instances
- sequence number - 12 bits - rolls over every 4096 per machine (with protection to avoid rollover in the same ms)
You should use NTP to keep your system clock accurate. It protects from non-monotonic clocks, i.e. clocks that run backwards.
- A standard PostgreSQL 12 or 11.4+ installation with development
environment (header files) (e.g.,
postgresql-server-dev-11package for Linux, Postgres.app for MacOS) - C compiler (e.g., gcc or clang)
git clone https://github.com/mausimag/pgflake.git
cd pgflake
make
sudo make install
You must configure the two following parameters before using:
pgflake.instance_id: Sets the id of current intance. It must be unique between instances.
pgflake.start_epoch: Sets the start epoch of current intance.
To activate extension, in a database you can simply do:
CREATE EXTENSION IF NOT EXISTS pgflake;
Create one table:
CREATE TABLE ttest (
x bigint,
y bigint NOT NULL DEFAULT pgflake_generate()
);
Now insert some data:
INSERT INTO ttest (x) SELECT generate_series(1, 5);
Check generated IDs:
SELECT * FROM ttest;
x | y
---+---------------------
1 | 1231425388031266816
2 | 1231425388610080768
3 | 1231425388626857984
4 | 1231425388631052288
5 | 1231425388635246592
(5 rows)
You can use the following functions to extract information of id:
SELECT
x
y,
pgflake_extract_time(y) as t, -- returns creation time of id
pgflake_extract_instance(y) as i, -- returns instance number
pgflake_extract_sequence(y) as s, -- returns the sequence number
FROM ttest;
