If you are really interested in this subject, you might want to read https://dataintensive.net/

If you clarify what your goals are, broadly speaking, maybe we can give advice.

Thanks a lot for the suggestion and the link!

I’m mainly building minikv to learn more about real-world distributed storage and consensus—and to see how far I can take it as a personal project. Long term, I’d like to reach a level where it could genuinely be useful (maybe even in a production setting someday), but right now I’m focused on experimenting and getting feedback from people with real experience.

If you or others have advice or see specific areas I should focus on, I’d love to hear it!

▲

esafak 4 hours ago | parent | next [-]

I think this field is mature, as far as it goes, so you'll be in the weeds by the time you get to the frontier. Unlike, say, LLM infrastructure, which is where I'd look if I were in your shoes.

▲

whispem 4 hours ago | parent [-]

Thanks for the insight!

You’re absolutely right—the distributed KV/object storage space is very mature, and I don’t expect to “out-innovate” the current leaders. My main goal is to learn by reimplementing challenging systems from (almost) scratch and to deeply understand how these pieces work behind the scenes.

LLM infra is definitely a hot frontier—makes sense that it’s where lots of cutting-edge work is happening. Maybe someday I’ll try my hand at that too. For now, I want to really master the fundamentals!

Thanks again for your perspective—it’s helpful to think about where the opportunities (and weeds!) are.

	▲	cannonpalms 3 hours ago \| parent [-]
		You're absolutely right(em dash) Hilarious

▲

praddlebus 4 hours ago | parent | prev [-]

> consensus—and

Even this guy's comments here are very obviously LLM generated.

▲

whispem 4 hours ago | parent [-]

No AI involved here—just me doing my best to be clear and thoughtful in my replies.

	▲	praddlebus 4 hours ago \| parent [-]
		LOL