โ NomadNet 1.4.2 released
๐ฌค rns.recipes
top
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Distributed KV store โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Started by never ยท 3d ago ยท f8dc375f83d40c1c
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
post-1
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ never #1 โ
โ f8dc375f83d40c1c โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โธ 3d ago
I've been thinking if there is place for a KV built on top of the RNS. The current trust
model seems to be more focused on trusting other nodes to keep your data (example: this
forum), host your own data or have custom solutions like lxmf propagation nodes. What if
we could have a distributed KV store to store arbitrary data? Is this somehow against the
Zen? I'm not entirely sure how a consistent hashing algorithm could be implemented in a
network that potentially is rebalancing all the time and how indefinite storage could be
achieved. LXMF does't give you guarantees but one would expect some sort of durability
from such kv. Has anyone been thinking about this problem? Is this just not correct way
to think about data ownership? Should we own and manage our own data (ie: contacts,
favourites, notes, TODO, etc) instead of offloading it to the network? Thoughts?
โ 3 โ 0 โค 2
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
post-2
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ steveplays #2 โ
โ daf042c1c000a24b โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โธ 3d ago
I also thought about doing data storage this way. I was thinking of something that
functioned as both a data store and a data history store (so a version control system).
Seems in line with the Zen to me, as long as node owners can manage what they mirror
themselves.
I was thinking of a repository simply being a collection of files (including history
files, containing a graph of changes), where the data hashes (folders, files or chunks)
can be signed with a Reticulum identity to ensure them being good. Of course, you still
need to trust the Reticulum identity/network identity that signed the hashes.
If hashes of file chunks are also tracked, nodes that mirror out of date files could also
still contribute by sending the up-to-date chunks (if not the entirety of the files
changed, which is often the case).
In my design, I was thinking that the data you want to mirror should be manually
specified (e.g. allowed folder/file/chunk hashes list or allowed repository hashes list).
I'm not sure if mirroring arbitrary data without opt-in is a good idea. If it's very easy
to offload data, it will get overused and become centralized towards whoever has the most
data storage, because others will not be able to keep up with the demand. Also seems like
it could allow for wasteful storage use. Either way, with my draft you could get around
this by making (and having someone allow) a repository (containing arbitrary data).
I'm not sure how to improve on this yet. What do you think, never?
โ 2 โ 0 โค 2
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
post-3
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ never #3 โ
โ f8dc375f83d40c1c โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โธ 3d ago
I was thinking more of generic data structure and less about files and folders but in
practice it would work the same. An immutable/append only data structure just like git
would have the benefit of being easy to synchronize and even guarantee some transactional
guarantees. I think that selective sync would be hard to manage. instead each node
operator would have to define a storage quota. in order to guarantee that data would not
be lost multiple copies would have to be stored across different nodes.
I understand the abuse concerns. I don't understand how this would be solved without
governance. even Blockchain is a form of governance with algorithms and economics. if you
think about today's git on rns implementation, it is basically what you are describing.
each user only hosts what they want, and as someone that wants to push data I have the
choice to push to multiple hosts.
The trust is done point to point. I choose to trust A to host my git and A trusts that I
won't abuse the storage.
โ 0 โ 0 โค 0
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
post-4
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ steveplays #4 โ
โ daf042c1c000a24b โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โธ 3d ago
rngit is slightly similar to what I have in mind, yes, but it's not the exact same. rngit
doesn't automatically pull from multiple mirrors, if it would be faster, for example.
Why would it be hard to manage selective sync?
It takes more effort to manage, sure, but we could figure out a way to make it easier to
use while retaining control over what you mirror. Similar to how network identities make
it easy to trust a group of identities attached to it and similar to blackhole list
sources.
โ 0 โ 0 โค 0
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
post-5
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ anon #5 โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โธ 3d ago
[freenet](https://www.youtube.com/watch?v=3RBNboYUlVI) explores a pretty similar part of
the design space to what you guys are talking about, any site or service that can be
reduced to database looksups can then be made decentralized, from github to reddit type
sites.
The only possible problem this decentralized nature would have would be higher network
overhead, but i don't think it would be an order of magnitude larger than what we have
already.
โ 1 โ 0 โค 0
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
post-6
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ steveplays #6 โ
โ daf042c1c000a24b โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โธ 3d ago
anon: I've seen the talk on freenet, but I wasn't fully impressed at the time. It was the
the presentation itself that made me a little disappointed.
Either way, probably a good resource indeed. IPFS is relevant as well and has been
mentioned on rns.recipes before. I took some inspiration from IPFS and Pijul in my design
draft. I was thinking of having commutative deltas like in Pijul for rns too.
โ 0 โ 0 โค 0
โ INFO โ Identify to this node to post. How?
bottom