Unique object identifier format with the possibility of lexicographic sorting.
A Go implementation of FOxID created by Kasefuchs and Daniel.
Here at LutraCorp, we spent a long time deciding between Snowflake and ULID, but each had its own disadvantages, and in the end we decided to create our own identifier format that includes the advantages of both. FOxID is a 128-bit identifier that is encoded using Crockford's Base32 and includes timestamp, generator identifier and two security measures against duplicate identifiers.
Go: >=1.22
$ go get -u github.com/lutracorp/foxid-goUsage examples can be found on GitHub. Reference pkg.go.dev.
Bit breakdown for an FOxID e.g. 065DTQHTA65T6JGMGBCTXT9P1M (counter is 8575406, random is 15283725, datacenter is 35747 and worker is 18964) looks like this:
000000011000101011011101010111100011101001010001 1000101110100011 0100101000010100 100000101101100110101110 111010010011011000001101
|------------------------| 24-bit random
|------------------------| 24-bit counter
|----------------| 16-bit worker id
|----------------| 16-bit datacenter id
|----------------+----------------| 32-bit generator id
|------------------------------------------------| 48-bit timestamp
UNIX-time in milliseconds since epoch.
- 48-bit unsigned integer.
Identifier of this generator.
Note You can combine datacenter and worker ids to get an even larger field which can contain 32-bit unsigned integers.
Identifier of datacenter on which FOxID generator is running.
- 16-bit unsigned integer.
Identifier of worker on which FOxID generator is running (e.g. process id).
- 16-bit unsigned integer.
Incrementing number to prevent ID conflict.
- 24-bit unsigned integer.
Random number to prevent ID conflict.
- 24-bit unsigned integer.
Each FOxID is a big endian byte buffer. Which can be represented as a string using Crockford's Base32 encoding.