There are a few useful features of having flexible routing. Each type has its
applications:

> FROM: finney.org
> DATE: 04/17/2000 09:55:27
> SUBJECT: RE:  [Freenet-dev] hwo do we want it?
>
>              Mike writes:
>              > The TOS flags that I would suggest would be:
>              > normal : (default) the way things are routed now

Normal routing has the advantage that it likely has the best chance of finding
something popular. I think that it does have some shortcomings in that where it 
finds
something may not be where it is the quickest to retrieve. Also, it is not 
going to
find something that someone is trying to keep alive on their node (information
analgous to a static web page). The argument could be made that this type of
information should be served through normal httpd means but freenet does still
provide a certain amount of anonymity for hosts of this nature.

>              > throughput : the node accesses the  node status table and 
> strongly
> biases the routing
>              > to nodes with high bandwidth

Max throughput routing would be useful in situations where large data files are 
being
accessed. The file may be popular and there is no doubt that it is out there
somewhere but you don't want to have to suck it through a thin straw. Providing
throughput based routing for requests (and maybe even inserts) will cause 
another
metric of key closeness based on data size ... this would be very useful to hurd
large data requests where they can be handled most effectively.

>
>              > speed : the node accesses the node status table and biases 
> routing
> to nodes with low
>              > ping times

Speed (or minimum delay) routing would be most useful in situation where the 
data you
are requesting is small and multipart. For instance, a future metadata searching
mechanism might return multiple small data chunks. Using this type of routing 
would
search through various data clusters grouped through the normal means and would
reduce the impact of the high percentage overhead inherent in requesting small 
pieces
of data.

>
>              > load : bias routing to nodes with low cpu load

This type of routing would enable the by-passing of congested areas and might
contribute to the overall health of the freenet. If a node gets overwhelmed with
requests, the load routing will self-regulate the flow to that node and 
distribute
the traffic a bit.

>              > random : just selects a node on the list at random (the 
> desparate,
> shotgun option to
>              > request data from those hard to reach places ... might get 
> data from
> that lone node
>              > that still holds some rare items)

This is the type of data request that adds a little chaos into the mix. This is 
good
for not letting data get too concentrated in a certain "area" on the freenet and
providing crosslinks between data clusters. Also, as I mentioned earlier, this 
type
of request might make otherwise unaccessible data available again. For example, 
data
that someone is keeping alive on their own node by re-requesting it over and 
over.
They can't re-insert it until it has expired from their node and requesting it 
keep
it local and the normal routing would not find it unless you put the htl up very
high.

>
>              As I wrote before, I don`t see how these features would allow 
> you to
>              find the data.  I am concerned that Freenet won`t work if you 
> start
>              changing the routing algorithm.
>
>              Hal

For what I can tell there are many advantages to having different routing 
techniques.
They built it into TCP/IP for a reason.

Mike


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

Reply via email to