Shameless plug: Anyone bothered about the wasted space in UUID string representation (and using Ruby) can check out https://github.com/sudhirj/shortuuid https:/

Shameless plug: Anyone bothered about the wasted space in UUID string representation (and using Ruby) can check out https://github.com/sudhirj/shortuuid https://github.com/sudhirj/shortuuid - it re-encodes your UUID into any alphabet you choose, with a Base62 default (I find that to be a sweet spot that gives both URL safety and efficiency). Let me know if you want ports in any other languages - the the algorithm is to really just treat the UUID as a hexadecimal number (that's actually what it is) and re-encode it into any other alphabet of choice. That said, always use native UUID types in datastores - they'll convert to bytes / numbers internally and will always be the most efficient. For other situations, remember that they're just numbers, so you can write them in binary, ternary, octal, decimal, hexadecimal, vowels, baseXX or really any other alphabet you want. The bigger your alphabet (as long encoding remains efficient, like ASCII under UTF-8), the better your gains will be.