The use of RIPng

[Jeff McAdams]

On 6/1/10 2:08 PM, bmanning at vacation.karoshi.com wrote:
> On Tue, Jun 01, 2010 at 06:29:51PM +0100, Nick Hilliard wrote:

>> Ok, let me spell it out.  If you're running a routing protocol on your
>> end-user workstations, you're probably doing it wrong.  If you're running a
>> routing protocol on your routers, then it goes like this:

> 	if you run a routing protocol on an end-station - you've turned it into
> 	a router...  (much like what happens in some anycast clusters for node
> 	failover)

Uhm, no.

If you run a routing protocol on an end-station - you've given that 
end-station a mechanism that it might learn what the network topology is 
in the overall network, beyond just its default next-hop.  You *might* 
let it be a router, depending on how that routing protocol is set up and 
other configuration issues within the OS.  ( echo 0 > 
/proc/sys/net/ipv4/ip_forward    pretty much makes a Linux box *not* be 
an IPv4 router, regardless of what software is running on it (yeah, 
yeah, unless you start getting into user-space routing and such))  You 
might also give that end-station the ability to inject routes into that 
network topology, which could, indeed, cause problems.

So, there are use cases where it could be beneficial for end-stations to 
have knowledge of the overall network topology by running a routing 
protocol.  There are also, almost certainly drawbacks.  I think it is 
possible for reasonable people to disagree (including based on their 
individual scenarios for use-case) on which is bigger, the benefits or 
the drawbacks.

-- 
Jeff McAdams
jeffm at iglou.com