Problem with IPFS

[Amitkumar11885]

Ok so I adore decentralization, and IPFS is great. The files once published will never go offline and because it is a peer-to-peer network, the load time for any file should be less in comparison to pulling a file from a server. However, it is not so. Even though the concept is great, the implementation is not meeting the expectations.

There are a few Core problems with IPFS. The first being that no user would want to setup a full node. People on the IPFS network now are tech enthusiasts, however, no user will want to deal with hashes. If a public node is available at ipfs.io/ipfs, why in the world should I setup my own node?? This would cause a shortage of nodes. I myself am guilty of joining the network, uploading my files, pinning them on public gateways, and leaving.

After that another problem is that what if I upload illegal content, and make a few nodes pin it?? IPFS, may be the replacement to torrents, which is not a good thing.

Next Filecoin. As great as it might be to get an incentive to store files, Filecoin will just drag IPFS down. If I have to pay to get back up my files, why should I not use Google Drive?? Over there At least my files are private and there is no risk of losing them.

Finally, hosting a website on IPFS. IPNS is simply terrible, and not user friendly at all, though that may change in the future. My main concern about hosting a site on IPFS is why should I do it?? There would be at max a few thousand nodes running on the network. Even if all of them pin my site, A CDN service such as cloud flare would be way more effective and efficient. Though it only works for static websites, IPFS can also only deal with static sites, as it is only a file system, not a compute sharing service.

IPFS also consumes a lot of bandwidth and users with metered connection don’t appreciate that.

Give me your thoughts on the matter.
ground Zero

[rayj001]

I have some of your same concerns and look forward to the replies from any IPFS gurus as I am totally new to IPFS.

[leerspace]

I don’t currently have the time to respond to every statement and question in your post, but here are a few points.

@Amitkumar11885

pinning them on public gateways

Are you sure they pinned it? Most (all?) major public gateways don’t expose their API to the public – so you can’t pin anything on them. It’s trivial to get them to cache files for you, but they are likely to be garbage collected.

@Amitkumar11885

After that another problem is that what if I upload illegal content, and make a few nodes pin it??

You can’t make other nodes pin your content by design – unless they’ve exposed their API to the internet for some reason.

@Amitkumar11885

IPNS is simply terrible, and not user friendly at all, though that may change in the future.

This same argument about user friendliness for the common internet user could be made about DNS. And that’s used all over the place today.

@Amitkumar11885

There would be at max a few thousand nodes running on the network

What are you basing your estimate on? I’m pretty sure there are more than this today, but Protocol Labs would have a better up to date count since they run the default bootstrap nodes.

@Amitkumar11885

The first being that no user would want to setup a full node.

Why not? It’s also possible for applications to run a full node in the background without the user even knowing IPFS is being used. The IPFS desktop application and Textile Photos for iOS are specific examples.

@Amitkumar11885

Even if all of them pin my site, A CDN service such as cloud flare would be way more effective and efficient.

Let’s say you have several neighbors with an IPFS peer caching a piece of content you want, and the nearest Cloudflare CDN server is in another country. I’m not sure how a CDN would be better in this example.

[hector]

I fully understand the sentiment here. We all want IPFS to be awesome enough to replace many of the centralized solutions that dominate now and we all find that it is not quite there yet. But you see, IPFS moves in a space that ranges from developing and implementing novel research in the p2p space to using existing p2p tech in a ruthless realworld environment to provide a generalist content-addressed filesystem (more versalite than, say, bittorrent, but also much younger). And some things here are quite tricky.

In that scenario I don’t find surprising that on several aspects IPFS cannot compete with traditional web tech which has had a much longer time and support to be optimized. Our team and community is however very committed to push things forward.

@Amitkumar11885

The files once published will never go offline and because it is a peer-to-peer network, the load time for any file should be less in comparison to pulling a file from a server. However, it is not so. Even though the concept is great, the implementation is not meeting the expectations.

The files will never go offline as long as someone cares about them. Otherwise they will go offline (pretty quickly). I also want IPFS to be faster than a central server, but a central server or CDN is purely optimizing for speed, while IPFS is optimizing to remove the central server. Compare how much faster and convenient is to use your VISA card vs. running a bitcoin node for the sake of payments. This does not mean IPFS cannot be faster (it is in some conditions), or will not be faster in the future (it will).

why in the world should I setup my own node??

Well, because you adore decentralization and you want to be a good citizen by strengthening the network. But you may have other reasons, like content-addressed data properties being useful to you or the fact that you don’t want to depend on a third party, or offline-first matters etc.

@leerspace

Even if all of them pin my site, A CDN service such as cloud flare would be way more effective and efficient.

Let’s say you have several neighbors with an IPFS peer caching a piece of content you want, and the nearest Cloudflare CDN server is in another country. I’m not sure how a CDN would be better in this example.

We do not choose peers to download from based on geographical locality (yet!). A best example is to imagine that you have lost connectivity to the Cloudflare CDN server but a peer in your LAN has the content. Resiliency and offline-first scenarios are key here.