How We Made IPFS Content Publishing 10x Faster
probelab.io
probelab.io
Making things faster by doing less (and not the same) been speeding up computing since forever! Can't help but feel like it's slightly misleading to call the providing ("publishing") faster when it's not actually doing the same, it's just that most parts turned async instead of waiting for confirmation.
Wouldn't this lead to the problem where the user things everything been provided properly, but once others try to find it, the records haven't yet been published? As far as I understand, it'd still take mostly the same amount of time until the entire CID (not just some of them) are available to others, the only thing that got "faster" is the end-user UX of the one providing?
That said:
>In practice, at least one of the 20 follow-up requests fails in the vast majority of operations, and a single unresponsive peer can stall the entire phase waiting for a timeout.
It continually surprises me how often systems lack a Fast Fallback-like strategy¹. Or at least sound like it. Just an absolute flood of apps and websites and systems that try to do something once and then never tried an alternate route until that finishes, something like a minute or two later... for a process that usually takes less than a second. It's maddening. By the time you're considering one to be "stalled" and delaying everything unnecessarily, you probably should've already started trying two or three alternate routes!
I felt the article addressed that a bit further down. 20 copies is a somewhat arbitrary knob in the Kademlia DHT design IPFS is based on and this lab's research suggested that 15 was probably closer to good enough for GET requests to succeed at about the same time cost. Rather than dropping the knob for the entire DHT, because redundancy is always useful in the long run they went with the Early Return and a secondary process called the Reprovide Sweep that still tries to push the network towards the 20 live copies minimum it desires.
I'm assuming the Reprovide Sweep was work previously done/documented because it seems like something that might have been more interesting to discuss at longer length in relevant parts of the article.
This is a probabilistic system anyway. Even if publication finishes to 20 nodes, why is that enough to return to the caller? Shouldn't it be 30, or 50, just in case?
I'd say it makes sense to return control once zero PUTs have been made and do the whole thing in the background, to avoid serializing operations that usually don't need to be serialized, such as publishing multiple objects.
I’m not talking about technology demos such as Wikipedia-on-IPFS (which indeed worked and was impressive) but where IPFS is actually being relied on for some functionality.
I’m not sure if it was actually a well thought out idea, since it’s not like fb has a lack of servers or bandwidth, but I didn’t work on it, so I don’t want to throw too much shade at it.
And tbh it shouldn't be terribly difficult to extend the existing infra to supporting a full IPFS based system but I don't think anybody has considered it worthwhile yet.
ATproto uses just the bits it immediately needs even if it could probably benefit from the other parts long term (ex for archival relay stream preservation).
It is a huge server traffic saver.
Used to use it to host different static websites on custom domains while Cloudflare's gateway was working, stopped using it for this purpose since its sunset in 2024.
Neocities used to publish their websites over IPFS, this feature got broken (for years) and finally got removed: https://github.com/neocities/neocities/issues/352
It used HLS with .ts files, in a special way which circumvented cloudflare protection against .mp4 and .ts files, or something along the lines. Don't remember the details, but it was a cheap way to deliver your stream to any video player using standard protocols only.
>The purpose of this list is to allow IPFS node operators (e.g. someone running a public IPFS gateway) to opt into not hosting previously flagged content.
IPFS node operators, who are supposedly interested in hosting malicious content (and i2p-hosted phishings are a real problem) can OPT INTO using this list.
In this case, I don't see how that's any problem for piracy - people can just use one of the bad/unfiltered nodes.
edit: Not opposed to the idea, just curious what makes you pick IPFS over the existing alternatives.
Most of the emulator frontends I've seen are pretty against integrating this kind of ease-of-piracy stuff, though, accepting recognizing and filling in metadata for well-known roms, but not making it easy to integrate with remote libraries of roms... except tools that run on "hacked" consoles, which seem to love just giving you a list of games with a "tap A/X to pirate" UI.
You can use fuse-btfs [0] for mounting torrents as filesystems! Last I checked it was a fairly mature piece of software so hopefully it doesn’t feel unnatural.
At any rate you can replicate the same thing by just hosting the ROMs on your own cloud storage and using something like macos virtual files which will do this transprent download/delete to manage storage.
You can make an archival copy, but it is copyright infringement to use that copy in an emulator.
For newer console generations (decrypted) backups are not legal due to needing DRM circumvention.
It basically works as BitTorrent, but also provides HTTP access to the files.
In fact, many pirate websites use IPFS in one way or another (either directly, by serving the downloads over one of the public gateway, or indirectly, for internal needs).
The main drawback of BitTorrent v1 file hashing scheme inside the .torrent file is that it makes a virtual stream of the directory you want to share, splits that stream into blocks, and hash it altogether. That means that each file inside the torrent is unique, even if your directory have very common files, each is which exist in BitTorrent network per se.
IPFS solves this issue by just hashing the contents of the file. The "directory" of the files is a list of each individual file. If somebody has created a directory of 100 mp3 files and you happen to have one of it and added it into your IPFS daemon, you will be serving this file even if you've downloaded it from elsewhere (not from this "directory").
2. IPFS has both immutable and mutable files and directories
In BitTorrent, if you want to update your torrent file (in Russian we use «раздача» (bittorrent "upload"/"seeding"), a very precise word for bittorrent which doesn't have direct equivalent in English, so I'll stick to "torrent file), all the swarm needs to be switched to the updated version due to #1.
This means you have split swarm: the majority of people still seed the old version of torrent, and the minority seed the new version. Because of non-content addressation, all the old, already exising files in new .torrent file are treated as "new", and the old swarm can't seed them.
In IPFS, you can either create new immutable directory with all the old files and 1 new file, and all the old files would be seeded by the existing peers, or you can create mutable directory, and you can just modify it to your like without the need to update the link.
---
Both of these issues are solved in BitTorrent v2 more or less, but it's still not very popular, even if the specification is from 2017.
IPFS however is much more featureful than that. It allows to build decentralized distributed (serverless) websites and services, with user data and such.
There used to be a ZeroNet project which directly aimed at decentralized distributed web services, and is was very cool, there were many blogs, forums, boards. It all could be implemented in IPFS, but I saw only very simple text editors over IPFS and such, much simpler applications than it was in ZeroNet.
BitTorrent had their own version of distributed websites, Maelstrom: https://www.ctrl.blog/entry/bittorrent-maelstrom.html
And this does exist! See confluence by anacrolix:
It seems pure HTTP tracker + Torrent is good enough.
sigh So entire IPFS + IPFS Gateway needs to be so huge? 10MB? My dot.exe (DVFS repo manager) is 150kB..
Also, ipfs directly fixes one of the bigger issues:
> Instead of storing the data on-chain, NFTs instead contain a URL that points to the data.
If it's ipfs, it points to the content. If it's ipns, it points to a changeable link to the content, but one that is made consistent through the network, preventing the trick of making it differ based on the referrer.
"If we do want to change our relationship to technology, I think we’d have to do it intentionally. My basic thoughts are roughly:
We should accept the premise that people will not run their own servers by designing systems that can distribute trust without having to distribute infrastructure. This means architecture that anticipates and accepts the inevitable outcome of relatively centralized client/server relationships, but uses cryptography (rather than infrastructure) to distribute trust. One of the surprising things to me about web3, despite being built on “crypto,” is how little cryptography seems to be involved!
We should try to reduce the burden of building software. At this point, software projects require an enormous amount of human effort. Even relatively simple apps require a group of people to sit in front of a computer for eight hours a day, every day, forever. This wasn’t always the case, and there was a time when 50 people working on a software project wasn’t considered a “small team.” As long as software requires such concerted energy and so much highly specialized human focus, I think it will have the tendency to serve the interests of the people sitting in that room every day rather than what we may consider our broader goals. I think changing our relationship to technology will probably require making software easier to create, but in my lifetime I’ve seen the opposite come to pass. Unfortunately, I think distributed systems have a tendency to exacerbate this trend by making things more complicated and more difficult, not less complicated and less difficult."
Not using it in production but i found it pretty cool to test
If also made the sites available on IPFS, but that was a secondary concern.
fine for publishing, but not for having a live data set that is both used and published at the same time, as you can do with torrent.
Maybe this is possible to implement using fuse/9p fs
Btw when we were working on our project HyveOS, we used Batman-advs routing table to quickly (really really quick) bootstrap new peers into the system.
Ah… sometimes i really miss working on this.
We need to get encrypted mesh networking (Yggdrasil, CJDNS, FIPS etc) out there before it’s too late.
[0] https://probelab.io/ipfs/dht/#chart-ipfs-dht-lookup-performa...
Its security posture was absolutely fucking gross the last time I reviewed it.
And of course, there's a shitcoin bolted on as well. Last thing I want to do is feed into FileCoin. Of course, everything new these days has some financial interaction crap bolted on to entice speculators and ilk.
Last I was told about it, there was no way to delete stuff from IPFS. Nothing enforceable, at least. Setting aside that public stuff is "impossible" to delete on the internet, there's something appealing to me about being able to shut off my server. Feels like that is less possible with IPFS hosted content.
Does anyone have some perspective for me about removing content?