Didn't get: How to recognize the protocol of different networks if we use p-p systems?.
One minute response: This is a problem, because most peer-to-peer protocols don't include a protocol field in their message format. Mostly, peer-to-peer systems rely on outside help to identify themselves. For example, a peer might assume that whatever host it's pointed at is running a peer of the same type, or maybe particular peers connect to particular ports, say all Freenet peers connect at port 10101.
Didn't get: How one can act as client and server both (practically).
One minute response: A client issues requests; a server responds to requests; a process can be both a client and a server by being able to both issue requests and respond to them.
Didn't get: In the peer-to-peer systems is it that there should be a directory? What exactly is a reverse engineering?.
One minute response: Yes, sort of. There has to be a directory, because otherwise peer-to-peer systems would be write-only storage. However, unlike other network-based storage systems, every component of a peer-to-peer system serves as part of the directory. Reverse engineering: somebody gives you a black box and tells you to figure out what it does and how it works without opening the box.
Didn't get: Like to read about peer-to-peer design.
One minute response: That's actually hard, because current peer-to-peer systems weren't developed under academic conditions, and so there was no need to obsessively document the details. It's fairly easy to get information on most current peer-to-peer systems, but there's little about the overall design principles that could apply to any peer-to-peer system.
This page last modified on 25 March 2003.