Filecoin: Proof of Storage Systems • Hacker News Board

Re: Filecoin: Proof of Storage Systems

#51

Post by Taek » Thu, Aug 20, 2020, 6:08 AM UTC

opportune wrote: post #47

One thing I'd like to see decentralized storage protocols guarantee is that not only does the data exist on the backend, but is it available enough to be useful, and if so, are the latency and bandwidth good. PoST might be able to solve this if it enforces a strict deadline on the answer and tests availability sufficiently frequently. The other is some sort of authentication. Encryption of the data itself, IMO, is no…

Can address this with probabilistic checks. Download random pieces of your data and ensure it has acceptable latencies. If not, move the data to a different provider.

By erasure coding your data over many providers, you minimize the amount of harm that a couple of bad apples can cause.

Re: Filecoin: Proof of Storage Systems

#52

Post by filleokus » Thu, Aug 20, 2020, 9:42 AM UTC

Dylan16807 wrote: post #46

filleokus wrote: post #31

I don't really get the point of decentralised _storage_ solutions, where the stated goal is to compete with ≈ S3 [0]. S3 storage pricing, for most users is just a rounding error compared to e.g egress or compute cost. And it has proven to be tremendously reliable, both in durability and access. No sane person would point a CDN against a Filecoin backed origin, right? For kinda-censorship resistant distribution of stu…

I understand your overall point, but when it comes to > S3 storage pricing, for most users is just a rounding error compared to e.g egress or compute cost. And it has proven to be tremendously reliable, both in durability and access. It's not like you can separate S3 storage costs from egress costs, though. I wish I could use S3 for storage but pay for my own bandwidth. Even if you'll to go all the way to their netwo…

Yeah, it's true, it's not like egress is a completely separate thing from storage.

But I guess it also depends a bit on your use case? E.g if you upload big datasets for heavy computation and then only download small results (≈ machine learning) or have a cheap(er) CDN in front of S3 with a favourable usage pattern, the storage and egress would be somewhat separate.

Still, something like Filecoin is obviously not a resonable solution for the run of the mill "cheaper than S3 file hosting and network" problem?

(I've been on the lookout for something like that, for personal use where I don't all those nines and other S3 benefits. The best thing yet seem to be Wasabi [0])

[0]: https://wasabi.com/cloud-storage-pricing/#three-info

Re: Filecoin: Proof of Storage Systems

#53

Post by hddherman » Sun, Aug 23, 2020, 8:25 PM UTC

fsflover wrote: post #17

simonebrunozzi wrote: post #2

> We created Filecoin because the amount of data humans generate is exponentially increasing, and we need more efficient ways to store and access it. I don't think Filecoin can be MORE efficient than some of the larger, current systems. I actually think this statement is misleading. It should say that Filecoin's goal is to offer storage options that are distributed (hence redundant), protected from censorship, and po…

It can be more efficient if you are using thousands/millions of hard drives which otherwise would just be unused.

A bit unrelated, but a pyramid scheme called Kairos had this exact same pitch. That didn't work out, but that was due to the fact that it was a scam, not due to any technical issues.