I have come up with a routing visualization diagram.

The legend for it is located at:
http://members.home.com/mwiktowy/freenet.legend.png

With all this talk of searching and update propagation and my inability
to explain myself in words. I remembered the old axiom that
"a picture is worth a thousand words" and put my internal visualization
tool
down onto virtual paper.
It is flexible, uncluttered for fairly complex routing situations.
It expresses the idea of key closeness, hops, node load, routing
failures, internode bandwidth, "regions" where closeness based routing
will be contained,
"sectors" where nodes know of one another, and many many more concepts
that
are sometimes difficult to put into words all in a diagram that kind of
looks like one of
my blood-shot eyes after a night of drinking heavily :]

I will offer a few words (referring to the legend above) to kick of your
understanding
of my scribble.

Closeness - the closer a node is to the centre of the circle, the closer
that a routed
hash is to the "ideal" node or "Focus" (at point F). The existence of a
Focus of course
depends on your choice of routing algorithm but the diagram is defined
at the
extreme that there is an IP that will match better than (or be closer
than) any other
IP to a particular hash. This provides a baseline to visualize other
routing techniques.

Regions - the circle at B - all routed messages within this, that are
routed closer
to the message hash, will remain in this circle (by definition) - this
is not a
representation of a region in physical space but rather in "closeness"
space.
This is the same concept as "Islands of Key closeness" that has been
expressed in
the past. Where two regions intersect, routing might get a bit chaotic.

Sectors - all nodes within a particular sector have references to one
another.
This has nothing to do with closeness but messages that are routed using
a closeness
metric will tend to stick within a tight sector shaped area on this
diagram.

Bandwidth - this can be represented with thicker lines for lots of
internode bandwidth.

Core - this inner region is where all nodes contain references to the
focus node. You will
note that, normally, routing only takes one hop to the focus in this
region since there
couldn't possibly be a closer node to route to.

Further routing diagram examples may be found at:

http://members.home.com/mwiktowy/freenet.examples.png

These examples illustrate the usability of the diagrams but do not
actually reflect
the behaviour of the current or future freenet.

Sources for these diagrams (after all, this is an open source project
;]) can be found at
http://members.home.com/mwiktowy/<filename>.sk - for "sketch" vector
drawing format or
http://members.home.com/mwiktowy/<filename>.svg - for "scalable vector
graphic" format

I hope this aids in everyone's (newbie and developer a like)
understanding of discussions of
future Freenet proposals.

Mike



_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to