Sagubooru Blog

So, it’s been 3 years since this project was first conceived, and while it may appear that things have been at a standstill for some time, trust me when I say things have not been stagnant behind the scenes. I’ve taken the main database offline, as Sagubooru is an old legacy now which is not likely used much. I’ve been working nonstop for the last two and a half years on a brand new architecture that accomplishes the original goals I set out to do when I first started this project, and much, much more. Those who were moderators, donators, or were well known should contact me over the upcoming months to take part in the launch for the first public testbed of the technology

A few tidbits about the technicals of the server:
100% C# 4 code base, ~25,000 lines, completely modular, Cross platform (Win, Linux, Mac, iPhone)
Fully encrypted node driven cloud (p2p with optional central server distribution assistance)
db4o driven database and encrypted blob data storage
Dynamic data objects with content tagging
Content Groups and Extensible Per-Object Permissions
Authoritative oversight over content by their respective owners and stewards
Guaranteed hole punching via UPnP, Stun, Stunt, Teredo, and a failsafe VPN
Event driven API and network communication
Three APIs available for client development:

• Discoverable WCF services (tcp, udp, http)
• Socket layer using an XML (binary planned) serialization protocol supporting both TCP and UDP.
• Directly hook as a library

To summarize:
You will be able to operate a node on the network, as well as connect to a central node or other nearby nodes (If you have permission to).
Both your node and your user account will have a unique permanent identity,
The network will still operate fully without the central server, however any changes to existing data on the network will be slow, as the object must be locked across all nodes in the swarm
You will be able to upload any content type to the network, and it will be accessible from other nodes even without the central server. Content can be private.
Fully user-driven network, stewards will only be around to handle the most egregious of issues. Spam is initially going to be an issue over the public feeds, however it should be short lived as the facilities are already in place to support the blacklisting of nodes and users from the network
Nodes are self updating (Strongly encrypted/signed), and security has been a major focus in the overall design. Binaries are signed, and a manipulated server will not be allowed to peer with other nodes in the main network. While the storage architecture has been designed to detect and resist any potential database poisoning, I don’t believe in overkill when it comes to security.
This will not be an anonymity network. While there is a substantial amount of anonymity implied by the decentralized design, it’s not a goal, and anonymity will never take priority over speed, reliability, or functionality. Initial testing has shown that this has the capability to be on-par with bittorrent for raw transfer speeds for established content, but it does not do so well for new content in extremely high demand (yet).

There is far too much to list at this point, and the initial version is currently feature complete and working. What is not yet complete however is the client interface. The “official” client is written in C#(WPF/SilverLight), and at the moment is not cross-platform. SWF and Ajax versions are entirely possible, however client development does not align with the commercial development of this framework, and so has to remain a second priority for a while. For those of who that may be interested in the code, this framework will likely be made opensource after initial public testing. The networking portion of this framework is known as the Atlas Project, and can be entirely detached and reused in other applications. This portion will likely be licensed under the GPL.