Re: [ANet-devel] Re: Basic requirements

[Dale T. <dal...@88...>]

I think we're getting mixed up here between the message protocol and the needs
of applications built upon it.  If you have a generic anonymous message library
that allows you to send a message between two nodes of the distributed network
then you'll be able to use it to send queries or data transfers.

I'm thinking about the mechanics of moving a piece of data around the network
no what the nodes do with the data once they have it.

The query is a broadcast message - please can I have a file called X.

Static data is a piece of a file - please transmit this data from A to B.

If you want a chat type system then you'll also need - please transmit this
data from here to this group of addresses.

thanks,

- Dale

On Fri, Dec 01, 2000 at 07:56:14PM -0500, Benoit Nadeau wrote:
> >Hi,
> >
> >Can you explain how you a query differs from a 'message' in the system?
> >Surely
> >a query is just another packet of data that needs to be communicated to
> >different node/nodes in the system?  I think that we are probably talking
> >about
> >the same thing.
> >
> >For this list I was focusing on the anonymous messaging part of the
> >application.
> >I'm am a bit confused about what you mean in section 2.2 of the documentation
> >about handling static data.  How would you propose to give the
> >functionality of
> >Freenet without using a similar system?
> >
> >What do you mean by static download (about Freenet)?
> >
> >thanks,
> >
> >- Dale
> 
> A query is like the searches in Gnutella: the data to transfer is small,
> and doesn't need to exists on the network for a long time. There is no
> check to see if a node already received a query. Thus, a node might receive
> duplicate queries.
> 
> For static data, I'll explain with an example. Let say you want to have
> newsgroups in ANet. Each message in the newsgroup should exist on the
> network for some days and the data is usually much bigger than a usual
> query (> 20 bytes). As all static data, each message will be marked with
> it's creation date (when it was sent), and a pseudo-unique key (128-bit
> random number). When there is some unused bandwidth, a node that has some
> static data will try to distribute it. But since the data is relatively
> big, the node will always ask its adjacent node if it already has the
> message. This check will make some extra trafic, but this might save a lot
> of traffic if the other node already has the data. Because static data is
> long-lived, the nodes will try to cache the data as long as they can (or
> want), thus most of the recent postings will be available from the network.
> 
> The big difference with FreeNet is that using static data is NOT
> recommended for file downloads, that is when a node wants to download some
> data from another node. This is because the "anonymous two way data
> transfer" allows any kind of download, while the transfer, in both
> direction, remains anonymous. Even using static data to store some files on
> the network for a long amount of time shouldn't be allowed. An example. I
> want to download the latest version of emacs (about 20MB, IIRC). If file
> downloads are made using static data, the 20MB would have to be
> "duplicated" to ALL the nodes on the network before the file reaches me.
> Unless half the nodes on the network would want exactly the same file, the
> precious network bandwidth will be wasted. A new version of emacs, and look
> at that network speed drop...
> 
> So, the difference between a query and static data is huge, because they
> are handled completely differently.
> 
> I'll put an example of a distributed network with my "passive file
> browsing" service.
> 
> - Benad
> 
> 
> _______________________________________________
> ANet-devel mailing list
> ANe...@li...
> http://lists.sourceforge.net/mailman/listinfo/anet-devel