Sia vs. IPFS

  • admins

    I don't follow IPFS super closely, though I'm moderately familiar with the project and try to stay up to date on all of the major milestones. If anyone seems something that appears incorrect, you should call me out on it because you might be right.

    IPFS is aiming to replace http. It's decentralized in the same way that the internet is decentralized: anyone can run a server and allow people to view content on the server. And if the server goes offline, that content becomes inaccessible. From what I understand, IPFS is also doing a lot of optimizations with caching, meaning that sometimes you can fetch files faster, but the caching is done on a volunteer basis and does not have a payment system attached to it. IPFS is good at doing content-based lookup. If you know the hash of a file, and anybody is serving that file over IPFS, you will be able to fetch the file from them. There is some level of routing that makes this possible, and it means you don't need to store a list of links to different servers or sites that may have your file, all you need is the file hash. On the other hand, there's no guarantee that someone will be serving that file over IPFS.

    Sia is quite a bit different. To get a file from Sia, you need to know both the hash and the person that has the file. In that regard, IPFS is better than Sia. But on Sia, the host is getting paid to store the file, and there's an incentive structure to make sure that the file does not disappear. IPFS provides no such guarantee, and big or expensive files are likely to disappear, and unpopular files especially are likely to disappear. On Sia, it doesn't matter how popular or unpopular a file is, it'll stay on the network.

    I think it's safe to say that IPFS and Sia are chasing completely different goals, and solving different problems. At some point in time, it may make sense to integrate them. The creators of IPFS for a while were also working on a project called Filecoin. They claim that they would like to come back and work on Filecoin again, though I the paper they released had several issues that make me think Filecoin needs to be refactored before it'll be useful. The goal of Filecoin is similar to the goals of Sia, but I don't really want to spend too much time talking about Filecoin until there's evidence that the project is being developed again. I do believe that Sia is a much better solution, and we've got at least 15 months of development that I do not think Filecoin has. Filecoin does use a new consensus mechanism, which means they'd likely need to do a lot of engineering work (potentially more than Sia) to get things working. I don't see Filecoin releasing a beta anytime soon.

    Sia is still too early to be looking at integration with IPFS. Even though the IPFS team seems to favor Filecoin, IPFS is fully open source and is a roughly decentralized protocol, which means it should be pretty easy to integrate Sia and IPFS even if their dev team isn't actively helping us. 'easy' could mean months of work, I'm not really sure what it would take to get the two working together. But it's on our list of things to explore when the core of Sia has more maturity.

  • Could it be possible to add IPFS's "I don't really need to know which person is holding the data as long as I have the hash" part to Sia without integrating everything else?
    That might contribute to solving the dynamic IP problem in a clean way, depending on how clean this "routing" is in IPFS since I don't know how it works.
    And even more depending on how it works, it might help with host semi-anonymity, which I care about...

  • I think the best part for Sia is that Sia is a marketplace for people buy and sell storage capacity. In this marketplace, There are incentives for people to store files.

    IPFS seems just technology and it is not a marketplace. To build a marketplace, it still needs to solve how to incentivize people to store the files.

Log in to reply