peer to peer - LeFreq/Singularity GitHub Wiki
Peer-to-peer networking is a cool concept that really exploits the full power of the Internet as a revolutionary tool. Rather than monolithic, client-server relationships, each peer or node is an equal in it`s ability to push (or pull) data.
Rather than:
Server
| ^ you have: peer <=> peer
v |
Client
It's a (perhaps) subtle, but significant difference. With modern computers, nearly everyone is a publisher of data, or certainly could be.
So this re-makes the internet. The Application layer and the Session layer in the OSI network model are inverted: the data administrator makes connections to other peers and this allows new routes to discovery and further understandings of the world.
In this OS, a peer standard of 4-7 nodes can be chosen, such that the node always tries to seek this connectivity and form ad-hoc, mesh networks. This gives access to millions of nodes in less than 10 hops.
Once a path is found, the packet should come back with the path, so that the network isn't deluged with seeking paths. Each path that returns successfully should be recorded. These paths are used for all subsequent packets, until a failure occurs.
Consider a sessionless p2p model, where all messages are single-purpose and no session is held open for dialog. Why? Router session support will make this obsolete. UDP only is more efficient, though...
Peers may be full-blown personal computers, high-end enterprise servers, or small point-of-sales terminals without a standard interface.
Consider these sources of data:
- Point-of-sale systems at business merchants
- HOP card transactions on transportation networks
- Fueling stations getting car data via Blue Tooth...?
- A stereo system getting radioID, CD in play. This data could be sent to a PC for data collection.
- A law enforcement DriverID input device to track motorist citations/stops.
Peers may be children within a larger peer node, controlling all data in and out.
"25 transactions from NODE POS1 ready to be received."
The supervisor node then can store these transactions as data for later reporting, while passign them onto the >> INTERNET at large for processing by banks.
SOME education on proper data stewardship must be given somewhere on the node. "While you may collect transaction and personally identifying data between the customer and your organization, you are not allowed to braodcast personally-idenentifying information beyond your organization. Only aggregate data may be shared for larger data colleciton and understanding."
Reporting largest customers by transactions, by dollars. Daily, weekly, monthly, and yearly sales. Time-of-day data. These should all have reports and some data categorization should be made to make this easy.
Q: Should a business be allowed to take customer name data, correlate it with an address from another source, and send mail addressed to customer?
Unlike case, a business may collect personally-identifying information from card transactions, for business reporting. They are resposbile for the proper handling of this data and are advised to keep it internally, except as they may hire trusted consultants to mind their data and give them reports... etc. They are legally responsible for the proper handling of this private data. Should you have reason to believe that this data is improperly handled, please contact Visa card services, etc. If you have unique transactions that single you out for discovery, it is up to you, the customer, to protect this activity, not the merchant. Even if you use cash, if someone knows something about your buying behaviors, you can be discovered. This is just the by-product of the information age and you will have to invent ways to obfuscate your activities if you wish to be unknowable. Merchants can advertise to customers that they guarantee that their transaction data does not leave their organization. CERTIFICATION.
Any transactions, no matter how small or seemingly insignificant has monetary value to the information universe, for good or evil. Singularity OS or third-party interest may develop standards for certifying an organization's data stewardship.
Visualization for CYBERSPACE:
- When a node comes online, it sends a broadcast to its peers, giving it's location and other possible data in the coordinate space if anything changed since last boot, otherwise only status. This is all sent with a single UDP packet, giving data on how much has changed and where to get it.
- Nodes can track the coordate and node data locally, looking only for changes to status (online or not) and position/color from other nodes, or seek updates from the broadcasts given above.
Special credit to Ian Clarke of the Freenet project for the ideas of data security through redundancy and peer storage.