Mirror of the Rel4tion website/wiki source, view at <http://rel4tion.org>
Clone
HTTPS:
git clone https://vervis.peers.community/repos/yEzqv
SSH:
git clone USERNAME@vervis.peers.community:yEzqv
Branches
Tags
verkosto.mdwn
Verkosto means “network” in Finnish.
Verkosto is a web service architecture. It’s not a single specific design, but an evolving project which will serve as the network layer of Razom. It will define how data and data management and access are done over the internet, and how client programs integrate with Razom through these interfaces.
Right now Verkosto is just an architecture. Later it will become a design and an implementation.
The goal is to have a distributed system which can function without servers, i.e. each participant in the network is a peer: Both a client and a server. Something like Bitcoin. At the moment, since I don’t have enough resources to focus on this, the architecture is based on a less distributed concept, in which there are web interfaces and storage nodes communicating over the internet. How federated queries are made is not yet specified.
[[TODO|TODO/OPEN]] explain the basics
[[DONE|TODO/CLOSED]] make diagram: [[diagram.dia]]
[[TODO|TODO/OPEN]] make the diagram convert to an image automatically in ikiwiki!!! I think SVG is best. Some plugins, especially pandoc (maybe txt and textile too?) can be a basis. See here. The image can be exported with Dia from a script, see man dia. Also see if the image generator plugins make the image as a file or embedded in HTML, because if it’s a file it can be reused on other pages, which is what I want.
[[TODO|TODO/OPEN]] bring all the text from my ideas page and from the paper
DHT
I need to start thinking about how knowledge would get shared and stored…
http://en.wikipedia.org/wiki/Distributed_hash_table
If a query just goes into the network, how do you decide how many peers to ask? You can’t ask all peers in the world for every single query, it doesn’t scale… I think. On ther other hand, how does YaCy decide whom to ask? Same thing. Needs research.
Another option is to divide the information into blocks and store them using DHT in the same way Freenet and GNUnet do. Then, a lot of data would be local on your system.
There’s also the idea of putting “who knows what” in the DHT, helping you find who you need to reach (that can use direct connection by IP).
Two tasks:
- Read thoroughly about DHT, understand precisely how it works and be able to make adaptations and improvements for my use
- See how others do it, considering their use case: Tox, YaCy, Freenet, BitTorrent, cjdns
With this knowledge, I’ll understand better.
IDEA: Basically my thing is like YaCy, but for semantic data instead of simple text search! Maybe I can just take it and improve.
Also read about how distributed SPARQL queries work.