Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

IPFS is the Distributed Web

ipfs.io

Post reply on HN
Print view View original

71–80 of 286 posts

Re: IPFS is the Distributed Web

#71

Post by viraptor »

empath75 wrote: post #67

As a devops guy, I sort of think ipfs seems more useful as a private, backend sort of solution where you trust all the nodes. I'm sort of vaguely imagining it running as a shared file system in AWS, running on docker containers.

> where you trust all the nodes

Why do you think that's needed? Of course if majority of nodes cheated and said "sure, I've got that file" and send you random garbage instead so you have to retry, that would be bad. But if majority are running proper implementation, you don't really need to trust anyone.

Re: IPFS is the Distributed Web

#72

Post by digi_owl »

idlewords wrote: post #9

I worry that this is another example of throwing technology at a social and political problem. That the current web is centralized has little to do with its technical design, and everything to do with economic and structural incentives that have made it that way. It's tempting to say "start afresh", but we'll just be trading our current problems for a new set of problems IPFS introduces. It's a law of nature that pro…

> I worry that this is another example of throwing technology at a social and political problem.

And here i was just listening to Adam Curtis talk about his latest creation, Hypernormalisation, where he touched on the issue of thinking one can solve social issues with engineering.

Re: IPFS is the Distributed Web

#73

Post by rmc »

runeks wrote: post #30
lukaslalinsky wrote: post #28

Earlier quoted context omitted.

If I get it right, the idea is that if you want to make sure it's there one year later, you set up your own server and host the files on your own. IPFS just allows others to distribute a mirror of the files.

Why would nodes in the IPFS network voluntarily act as a load balancer for my VPS? Usually I have to pay for load balancing, so I'm a bit suspicious when someone claims I can get it for free from IPFS.

> Why would nodes in the IPFS network voluntarily act as a load balancer for my VPS?

(a) People currently downloading and viewing the site. Like bittorrent, the more popular a site/page is, the more people will be seeding it. (b) The Internet Archive.

Re: IPFS is the Distributed Web

#74

Post by kirushik »

viraptor wrote: post #71
empath75 wrote: post #67

As a devops guy, I sort of think ipfs seems more useful as a private, backend sort of solution where you trust all the nodes. I'm sort of vaguely imagining it running as a shared file system in AWS, running on docker containers.

> where you trust all the nodes Why do you think that's needed? Of course if majority of nodes cheated and said "sure, I've got that file" and send you random garbage instead so you have to retry, that would be bad. But if majority are running proper implementation, you don't really need to trust anyone.

IPFS is content-addressed. That means that it's trivial to discard tampered content. Lying majority, sure, will impact your content retrieval speed — but dealing with cheaters seems to be as simple as adding their addresses in some ignore list (which can itself be IPFS-shared, if needed).

I'm not sure of this blacklisting is implemented in current go-ipfs out js-ipfs codebase — but hash verification itself certainly is.

Re: IPFS is the Distributed Web

#75

Post by diggan »

tijs14tijs wrote: post #49

Interesting, I have two questions: Can you create your private ipfs network? (accessible by anyone, upload only me) If you upload sensitive material to the global ipfs network, what do you think will happen?

Private networks is something that we're working on (track it here: https://github.com/ipfs/notes/issues/146) but it would be private-private, only access+upload between your nodes. "Accessible by anyone - upload only by you" is not gonna be supported since it doesn't really make sense. If someone has a hash of a file, they can rehost it because they can access it. In IPFS you're not locating files based on WHERE they are but their name rather.

Re: IPFS is the Distributed Web

#76

Post by diggan »

fosh wrote: post #10

How do hosting providers fit in here, if at all? E.g., if I want to host a website on IPFS, do I publish it from my own machine and then wait a healthy amount of time for the content to be absorbed by the ether, or is there some way I can encourage other nodes to pick it up without requiring end-users to actively seek out my fresh material?

So if you add a file to IPFS, it's not gonna be automatically picked up by the network. Only people who request it, will have the file.

In the future, there will be hosting providers that allows you to upload a file to their IPFS "cloud" either as free hosting or paying for it, just as today. It'll just be easier for everyone to pass around a hash of the file rather than the file itself.

Maybe there will even be an API that you can hook up into your application, so content will be seeded that way.

Re: IPFS is the Distributed Web

#77

Post by diggan »

tscs37 wrote: post #13

IPFS is a pretty nice project, but it's pretty slow at times.

I work with IPFS so it would be very handy to know what you feel is slow exactly. We're working our hardest to make it as fast as possible but sometimes we fall through, so more information so we can fix it would be wonderful!

Re: IPFS is the Distributed Web

#78

Post by Kubuxu »

Kubuxu wrote: post #56
tscs37 wrote: post #16

Earlier quoted context omitted.

Well, as I understood IPNS is a bit limited; it only has one resource and there is only one authority that can change content, the node in this case. It does protect against DDOS, since old content remains visible and online but if the private key is leaked or the node is malicious, it has exactly 0 protection. Even worse, if the private key is stolen, you can't do much about it unless you have DNS setup... which bri…

IPNS is bug Proof of Concept that mutable content is possible in IPFS, there is long running goal of InterPlanetary Record System that would include much more complex validity schemes including revokeation.

s/bug/big/ in my previous comment

Re: IPFS is the Distributed Web

#79

Post by Kubuxu »

viraptor wrote: post #71
empath75 wrote: post #67

As a devops guy, I sort of think ipfs seems more useful as a private, backend sort of solution where you trust all the nodes. I'm sort of vaguely imagining it running as a shared file system in AWS, running on docker containers.

> where you trust all the nodes Why do you think that's needed? Of course if majority of nodes cheated and said "sure, I've got that file" and send you random garbage instead so you have to retry, that would be bad. But if majority are running proper implementation, you don't really need to trust anyone.

You need just one node to send you the right data. No need for the majority.

Re: IPFS is the Distributed Web

#80

Post by tscs37 »

diggan wrote: post #77
tscs37 wrote: post #13

IPFS is a pretty nice project, but it's pretty slow at times.

I work with IPFS so it would be very handy to know what you feel is slow exactly. We're working our hardest to make it as fast as possible but sometimes we fall through, so more information so we can fix it would be wonderful!

Most "slowness" is related to content that is either relatively fresh on the network or or rare content.

If my node is the only one seeding the file, accessing it via gateway.ipfs.io is slow but it's pretty obvious why; I'm the only provider.

So while for very public content IPFS is rather fast, if you share a couple files with a few friends, it's a bit slower.

Of course, this all depends on upload/download speeds, mine aren't stellar in both directions either.

Plus I'm rather certain that my ISP is throttling any kind of P2P traffic.

Post reply on HN